The Light Fantastic: Using airborne lidar in

The Light Fantastic
On 1st April 2015 the Historic Buildings and Monuments
Commission for England changed its common name
from English Heritage to Historic England. We are now
re-branding all our documents.
Although this document refers to English Heritage, it is
still the Commission's current advice and guidance and
will in due course be re-branded as Historic England.
Please see our website for up to date contact information, and further
advice.
We welcome feedback to help improve this document, which will be
periodically revised. Please email comments
to [email protected]
We are the government's expert advisory service for England's historic environment.
We give constructive advice to local authorities, owners and the public. We champion
historic places helping people to understand, value and care for them, now and for the
future.
HistoricEngland.org.uk/advice
2010
The Light Fantastic
Using airborne lidar in archaeological survey
1
Contents
Preface
Part III How do you use it?
Part I What is lidar and what does it do?
1
Visualisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1 What is lidar? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1 .1 Airborne lidar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2
2 .1
2 .2
2 .3
Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Archaeological . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Artefacts and issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Field use: hard copy, digital and vector . . . . . . . . . . . . . . . 27
2 What does it provide? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 .1 Height data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 .2 Intensity data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3
3 .1
3 .2
3 .3
Data types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Raw and gridded data; TINs and raster . . . . . . . . . . . . . . . . 8
Surfaces DEM, DTM and DSM . . . . . . . . . . . . . . . . . . . . 11
File formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4
4
Accuracy and resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1
Stonehenge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2
Witham Valley . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3
Forest of Dean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4
Mendip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5
Savernake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Part II How to decide if you need it:
practicalities and limitations
1 Project planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1 .1 MoRPHE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1 .2 Survey considerations and outputs . . . . . . . . . . . . . . . . . . . 17
2
24
24
25
26
2
2 .1
2 .2
2 .3
2 .4
Where can you use it? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Grassland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Moorland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Arable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Woodland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
To map or not to map? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4
Data acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5
5 .1
5 .2
5 .3
Dissemination, archiving and copyright . . . . . . . . . . . . . . .
Dissemination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Archiving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Copyright . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18
18
18
18
19
21
21
21
21
Part IV Case studies
Part V Lidar for woodland survey
1
Survey suitability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2
Identifying features in woodland . . . . . . . . . . . . . . . . . . . . . 34
3
Lidar and managing the historic environment
in woodland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Conclusion and summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Decision Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Appendix Sources of advice on using lidar . . . . . . . . . . . . . . . .
English Heritage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Forestry Commission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
39
40
41
43
43
43
44
Preface
These guidelines are designed to help
those intending to use airborne laser
scanning (ALS), also known as lidar, for
archaeological survey . The aim is to help
archaeologists, researchers and those who
manage the historic environment decide
first whether using lidar data will actually
be beneficial in terms of their research
aims and then how it can be used most
effectively . The guidelines will be most
useful to those who have access to data
that have already been commissioned,
or are planning to commission it for a
specific purpose . They also provide an
introduction to the interpretation of the
data to separate archaeological and nonarchaeological features .
Although the main themes will be
introduced, these guidelines are not
intended as a definitive explanation of
the technique or the complexities of
acquisition and processing of the raw
data, particularly as this is a still
developing technology . This document
is intended to be complementary to the
Heritage3D guidelines, which cover the
wider range of uses of laser scanning
for heritage purposes .
Part I
What is lidar and what does it do?
1 What is lidar? A brief history and
introduction to technology
Lidar, like radar, is an acronym and stands
for light detection and ranging, which
describes the method of determining
three-dimensional data points by the
application of a laser . It is a Remote
Sensing technique, using either groundbased (Terrestrial Laser Scanning (TLS))
or airborne systems (Airborne Laser
Scanning (ALS)), and is also referred to as
Airborne Laser Swath Mapping (ALSM);
in some military contexts it is known as
LaDAR (Laser detection and ranging) . For
the purposes of this guidance note the term
lidar will be used .
In its broadest sense lidar refers to a
much wider spectrum of techniques than
can be addressed in this note, and can
be used from static or moving platforms
including aircraft and vehicle mounted
scanners . It is the application of aerial
systems to which these guidelines refer .
As with many technologies that have
since been turned to uses within the
commercial and domestic sphere, lidar
had its origins in the military . Although
some of the first uses for laser beams
tested by the military utilised high-power
beams in attempts to destroy missiles etc,
the modern concept of lidar grew from a
somewhat less destructive idea . Following
on from the concepts behind radar,
Airborne Laser (or lidar) Bathymetry
(ALB) grew out of efforts in the mid
1960s to use the newly invented laser to
find submarines . Tests were carried out
in the 1970s and by the 1980s there were
operating systems in the US, Canada and
Australia . Current models of bathymetric
lidar, specifically the US Geological
Survey SHOALS system (Scanning
Hydrographical Operational Airborne
Lidar Survey), can map topography above
and below the surface of the water, down
to a depth that is roughly equivalent
to three times the visible depth . While
bathymetric lidar was being developed, the
concept was also extended to topographic
lidar for measuring surfaces on land .
One of the key reasons behind the delay
in implementation lies in the nature of the
lidar technique . As is explained in greater
detail below, the core constituent of the
lidar system only measures relative position
by recording the time taken for a single
laser pulse to be fired from a sensor array,
strike the surface of an object below and
be detected as a reflected signal . It is only
possible to calculate the actual location on
the ground by knowing the exact position
of the sensor array at the time that it fires
and records the beam . Previously, from
the 1960s, Transit (or NAVSAT – Navy
Navigation Satellite System), the world’s
first operational satellite navigation
system, had enabled positional accuracy
of c 200m, but this was insufficient
for the purposes of lidar . The launch
of the Navstar System, together with
improvements in the development of IMUs
(Inertial Measurement Unit, which records
the pitch, roll and yaw of an aircraft),
introduced accuracy of a sufficient level to
make lidar a practical reality . Early systems
were developed at NASA and commercial
models became available in the mid-1990s .
In this country the Environment
Agency (EA) began using topographic lidar
shortly after it became available with their
first surveys carried out south of Coventry
in December 1996 . Mapping began in
earnest in 1998 when they surveyed c 3000
km2 and has continued since that date .
The Environment Agency has used the
lidar data for the production of cost-effective
terrain maps suitable for assessing flood risk .
While their standard 2m resolution data
(one data point for each 2m2), an example of
which is shown in Figure 1, were the norm
until recent years and was entirely adequate
for measuring large-scale topographic
changes for flood modelling etc, it was
generally considered that this resolution
would not be suitable for the identification
of a wide range of archaeological features .
This assumption was based on previous
experience in examining satellite imagery at
a similar resolution . In fact, prior to 2000
it seems that the archaeological community
in the UK had not even considered the
possibility of using lidar for archaeological
survey and indeed very few had even heard
of the technique .
Lidar is currently used in a wide
range of scientific applications such as the
detection of atmospheric constituents .
One of the many general descriptions of
what you can do with lidar available on
the web is given by M7 Technologies, a
company that carries out research into
laser-based measuring techniques, and
shows the breadth of its interpretation;
it states that lidar can ‘measure distance,
speed, rotation, or chemical composition
and concentration of a remote target where
the target can be a clearly defined object,
such as a vehicle, or a diffuse object such
as a smoke plume or clouds’ (http://www .
m7tek .com/terminology .htm) . Elsewhere
on the web various sources report that
there are three basic types of information
that can be obtained:
3
Fig 1 The Roman fort at Newton Kyme, North Yorkshire showing as a slight earthwork (© Environment Agency copyright 2008. All rights reserved).
l
l
l
range to target (Topographic Lidar, or
Laser Altimetry)
chemical properties of target
(Differential Absorption Lidar)
velocity of target (Doppler Lidar)
System (GPS) and the Inertial
Measurement Unit (IMU) in the aircraft
(Fig 2) . Using the principle of measuring
distance through the time taken for a pulse
of light to reach the target and return it is
possible to record the location of points
Differential absorption will be briefly
covered under the heading of Intensity
Data, but otherwise these guidelines mainly
relate to the use of the topographic data
recorded by lidar and specifically those
from an airborne platform . It should be
noted, however, that the development of
mobile ground-based platforms may have
potential for the recording of earthworks
in pasture such as deserted settlements;
for small areas a ground-based survey is
likely to be considerably cheaper than an
airborne one .
1.1 Airborne lidar
In basic terms airborne lidar consists of
an active laser beam being transmitted
in pulses from a fixed wing or rotary
aircraft and the returning reflection being
measured . The precise location of the
sensor array is known due to a
combination of Global Positioning
4
Fig 2 Principals of lidar (after Holden 2002).
on the ground with a very high degree of
accuracy, typically 100–150mm in both
plan and height .
The majority of laser sensors operate by
sending out a laser beam that scans across
the ground surface by means of a mirror
(rotating or oscillating depending on the
sensor) or alternatively by a fibre optic
scanner . Whatever is the means of emitting
the beam, the calculations that enable the
creation of Digital Terrain Models (DTMs)
etc are based on the returning (reflected)
pulse to the sensor . In general, most
airborne lidar uses eye-safe lasers within
wavelengths in the infrared (IR) range;
those systems on the current market range
from 900nm to 1,550nm . The exception to
this is bathymetric lidar, which uses a twin
beam system; the green beam penetrates
the water and detects the seabed, while the
infrared beam detects land and sea surface .
Airborne lidar, therefore, provides
the ability to collect very large quantities
of high precision three-dimensional
measurements in a short time . This
facilitates very detailed analysis of a single
site, or data capture of entire landscapes .
It does not necessarily provide any
information about the point being recorded
in the way that multi-spectral data can, nor
does it give any inherent information about
the nature of the feature being recorded
(though see below for full waveform lidar) .
What it records is the three-dimensional
location of a point in space (together with
some information on the intensity of the
reflection) .
It should also be noted that unlike
some remote sensing tools lidar is an active
sensor in that it sends out a beam and as
such it is possible to use it at night or in
circumstances when passive sensors would
not work . It should be noted, however,
when planning a survey that flying at night
means that the aircrew are less able to see
whether there are clouds present that may
affect the quality of the survey, until after
the data have been processed .
For further details of the principles
behind lidar see Holden et al 2002, Pfeifer
and Briese 2007 or Wehr and Lohr 1999;
and for further information on the use of
intensity data see Challis et al 2006, and
Höfle and Pfeifer 2007 .
is described as being able to ‘see through
trees’ . This is a misleading statement,
however, and can lead to disappointment
if the properties of lidar are not properly
understood . The key element of lidar is
light, and as such it cannot see through
trees or anything else . However, in some
appropriate circumstances significant
gaps in the canopy can make it possible
to record the ground surface under
woodland, something that is discussed in
further detail below . What lidar will do
is provide accurate locational and height
data, enabling the creation of a threedimensional model of the land surface
that can be examined to identify historic
features that exhibit some form of surface
topographic expression, although this does
depend on the resolution of the data and
on other factors described in detail below .
The intensity of the reflection of the laser
pulse can also in some circumstances
provide useful information .
Like any other tool for archaeological
recording lidar has its strengths and its
weaknesses and it depends to a large
extent on the ability of the user to interpret
the data effectively . Lidar will not make
other techniques redundant, but will
rather provide an additional source of
data . Specifically because of the generally
relatively low resolution of the data (see
section I .4 for exceptions) airborne lidar
is best fitted to large area survey such as
is categorised as English Heritage Level
2 survey . Details of the different levels
of survey defined by English Heritage
are given in the guidance document
on understanding the archaeology of
landscapes (English Heritage 2007) and
should be considered before the initiation
of any survey .
2.1 Height data
There is a long tradition of archaeologists
interpreting historic sites from humps and
bumps visible on the ground or from the
air . However, the height data recorded
by lidar (as shown in Fig 5 below) is not
a straightforward record of the ground
surface . When the laser is fired from the
plane it travels towards the ground and
if it strikes anything in passing, part of
that beam is reflected back to the sensor
and forms the first return; the rest of the
beam continues towards the ground and
may strike other features that produce
further returns until it finally strikes the
ground, or a surface that allows no further
progression . The final reflection that
reaches the sensor is known as the last
return . In practice, built-up areas and open
land act as solid surfaces and the first and
last returns are often identical . Woodland,
however, functions as a porous surface
where the first return generally represents
the top of the tree canopy while the last
return may be a reflection from the ground
surface, but equally may be from the main
trunks of the trees or areas of dense canopy
or undergrowth (Fig 3) .
For many early-generation sensors
only a small number of return echoes were
collected from each pulse – often just the
first and last return, with occasionally an
additional one or two in between . The first
Summary
l For archaeologists the key value of
lidar is in providing accurate threedimensional measurements of a surface .
l Although lidar can be used from
stationary or ground based platforms,
these guidelines deal only with
airborne lidar .
2 What does it provide?
Lidar is seen by some as a tool that
will record all aspects of the historic
environment, making other techniques
redundant; this is especially true when it
Fig 3 First and last returns: the image shows the scatter of points returned by the laser pulse; the blue points represent the
last returns which have penetrated through to the ground while the red and orange represent those that struck the canopy.
5
Fig 4 Full waveform lidar (after Doneus) – The image shows how the full waveform of the lidar pulse is recorded over various ground surfaces.
and last returns were considered the most
important: the first being equivalent to the
Digital Surface Model (DSM) and the last
being used as a means to help calculate a
Digital Terrain Model (DTM) .
Within the last few years the latest
development of lidar sensors has expanded
and now, instead of just recording
between two and four returns, the
new full waveform system digitises the
entire analogue echo waveform for each
emitted laser beam (Fig 4) . During postprocessing, it is possible, by combining
the added detail from the whole pulse
of the beam such as the echo width and
amplitude, to produce much more accurate
models of the ground surface by more
accurately eliminating ground cover such
as low-level undergrowth, which can give a
false reading that appears to be the ground
surface (Doneus and Briese 2006; Doneus
Fig 5 Generic lidar tile showing heights differentiated by colour shading (lidar © Mendip Hills AONB; source, Cambridge University ULM (April 2006)).
6
et al 2008) . Being able to analyse the entire
waveform also means that it is possible
to obtain data from weaker returns and
achieve a more accurate observation by
better resolving the return information .
This technology is still partially in the
development phase; full waveform scanning
is a practical working product in so far as
there are full waveform scanners that will
provide good results (eg IGI LiteMapper
5600, TopEye Mk II and various sensors
from Riegl LMS and TopoSys GmbH),
although there is currently only limited
availability within the United Kingdom .
Furthermore, at the moment, there is
very limited consumer software that
provides full control over the analysis of
full waveform data (eg extracting echoes
from the waveform etc) on the market .
There is the standard operating software
that comes with the scanners, but this
tends to be expensive and you need to be
experienced to use it . Rather, it is currently
necessary to have contacts with those
institutions researching the topic . There
will, however, certainly be further advances
in full waveform analysis within the next
few years .
Although using full waveform
digitisation produces significantly greater
amounts of data at the time of survey,
after processing the size of the key
dataset, the DTM, is solely dependent
on the resolution required . Because of
the additional time and cost required in
production of the data use of full waveform
data may only be appropriate for vegetated
areas where the additional data can inform
and enhance the vegetation removal
processing .
However it is generated, the most
useful product of lidar for archaeologists
is the three-dimensional model of the
ground, the DTM, because of the
information it can provide in woodland; in
non-wooded areas the DSM is preferable
because of the absence of smoothing effects
(see below) . The DTM still requires careful
manipulation using specialist software, to
facilitate analysis and interpretation of the
archaeological features, discussed further
below (Fig 5) .
2.2 Intensity data
While the height data are generally
seen as the core product from the lidar
survey they are not the only information
recorded . As well as the relative x, y and z
position of the point on the ‘ground’ the
sensor also records the intensity of the
reflected signal . This can be affected by a
combination of factors (eg flying height,
laser power, atmosphere, direction of laser
Fig 6 Generic lidar tile showing the intensity of the returned signal (lidar © Forestry Commission; source, Cambridge
University ULM (May 2006)).
beam, number of returns), but as long as
these constants are known the data can be
calibrated such that the results are largely
determined by the wavelength of the laser
beam and the nature of the surface from
which the pulse is reflected; different
surfaces provide a different absorption rate
and consequently reflect back differing
signal strengths, which can be analysed to
characterise different surfaces .
As a result of this, after appropriate
correction, (Höfle and Pfeifer 2007;
Boyd and Hill 2007) the intensity data
can be used to analyse the reflectivity
of the surface being hit by the laser and
thus aid in interpretation . When seen as a
simple image file the intensity information
translates into a series of tonal differences
and provides an image of the return surface
similar to that of a true panchromatic
orthophoto at the same resolution (Fig 6) .
One point that needs to be borne in mind
is that because the lidar pulse is generally
in the near infrared (NIR) rather than in
the visible spectrum, the reflectance is not
what might be expected by those unused
to working with wavelengths outside the
visible range (eg those used to dealing with
standard aerial photographs) . Whereas a
flat, solid surface such as stone or concrete
will reflect almost all of the light in the
visible spectrum, this is not the case with
infrared light; instead asphalt for roads has
a low return value, while grass or a green
crop will have a high return
There may be archaeological potential
in using intensity values as a method of
assessing the moisture content of exposed
soils . A project funded by the Aggregates
Levy Sustainability Fund (ALSF)
investigated whether this could be used to
predict the likelihood of preservation of
waterlogged archaeological remains, but
results have proved inconclusive (Challis
et al 2008a) . While the results suggested
that from a visual standpoint the lidar
intensity data proved useful in qualitative
analyses of certain areas, the report stated
that ‘the application of lidar intensity
data to predictively model sediment
units of high preservation potential can
be deemed at present to be untenable’ .
However, while the usefulness of the
intensity data to identify damp ground
seems to be uncertain there is definitely
useful information in the data in other
circumstances .
Chlorophyll in plants reflects near
infrared radiation, so changes in the
chlorophyll content of a single plant
species, perhaps as a result of stress such as
drought, can be represented in the intensity
data in the same way that they are seen in
the visible spectrum as cropmarks . In fact,
because chlorophyll reflects c 50% of NIR
radiation, as opposed to 15% of visible,
plant stress (eg grass growing over buried
walls) is much easier to discern than in the
visible spectrum (Verhoeven and Loenders
2006) . This is something that has long been
recognised by archaeologists and was first
systematically investigated by Hampton in
the summer of 1970 when he reported that
compared to standard film the NIR film
‘showed distinct advantages at the early
stages of cereal growth’ (Hampton 1974) .
7
Fig 7 Lidar tile over Savernake forest showing the Roman road appearing as a feature due to the difference of the intensity
of the returned signal (lidar © Forestry Commission; source, Cambridge University ULM (May 2006)).
3 Data types
During the process of a lidar survey
there are a number of stages at which
data are generated and can be provided
to a client . However, in order to be able
to reprocess and manipulate the data to
gain the maximum benefit from them,
it is important to ensure that the most
appropriate type of data is chosen . It is
also important to be aware of the stages of
processing the data have been put through,
as these can result in data artefacts that can
be misleading .
The primary data are collected by the
scanner simply as a series of points in space
based on the calculation of the time taken
for the beam to return to the sensor . It is
only after these data have been registered
(placed in a common coordinate system)
that they are readily usable . This procedure
is carried out by the data provider . After
the data have been registered it is then
necessary to align the grids of individual
survey swathes to ensure that there are no
discrepancies between scans that could
lead to interference patterns . Again this
procedure is best carried out by the data
provider . These processed data can finally
be manipulated by the archaeologist
within specialist software to emphasise
the features of interest . Various different
software packages are used to produce
these processed data, but these software
packages are too varied for discussion here .
It is important to note, however, that users
should try to gain some understanding of
the processing that has been carried out by
8
their data provider in order to understand
any issues of data degradation or artefact
creation that may have occurred . This
is particularly important where filtered
bare earth DTMs are provided that may
have utilised classification algorithms to
extract and remove buildings and any other
features (see Part III 2 .2) .
The data can be provided in a variety of
forms and as a range of products (eg point
clouds, pulse data, images, DTMs, DEMs),
the suitability of which depends on their
planned use . It should be noted that
while there are a number of proprietary
formats in which laser scanned data can
be provided there is a growing general
consensus that the standard format for
recording the three-dimensional point
data should be the ASPRS LAS format V2
(Graham 2007) .
Unfortunately, because the use of lidar
within the archaeological world is relatively
new, the discussion of formats etc is quite
jargon heavy . Many of the terms used
will be familiar to those used to working
within GIS or certain other remote-sensing
techniques, but may be confusing to others
who are planning to commission a survey
or utilise existing data . While it is not
essential to understand all the technicalities
of how lidar operates – some of which
has been outlined above – it is useful to
understand the difference between the
different products .
3.1 Raw and gridded data; TINs and raster
The two most obvious differences are
between what are often referred to as
‘raw’ and ‘gridded’ data . In ‘raw’ data
the individual points are scattered across
the survey area exactly as they have been
recorded, while in ‘gridded’ data the survey
points have been processed to form a
regularly spaced array .
In the most basic of terms, raw data are
simply a series of tables that record the x,
y, z and intensity data for large numbers
of points on the ground (NB ‘the ground’
refers to the surface struck by the laser
pulse and does not necessarily equate to
a point at ground level) . If point data are
viewed as a text file they are simply strings
of numbers with columns for x, y, z and
intensity data values . Additional sets of
columns may be provided to separate first
and last (or even intermediate returns) .
Each row equates to data from a single
laser pulse .
The x and y points are calculated to
map the actual centre point of the laser
footprint (see Glossary) transformed to the
OS NGR and the z coordinates are the
elevations of the points of reflection . In
Fig 8 A point cloud showing how the general structure of features can be revealed (lidar © Mendip Hills AONB; source, Cambridge University ULM (April 2006)).
some cases the z coordinates are recorded
in centimetres or millimetres rather than in
metres, and this can cause problems when
plotting in GIS . When imported into a GIS
package these x, y and z points produce a
point cloud, which is exactly what it sounds
like: a cloud of points .
A useful way to imagine the result of a
point cloud, suggested by Peter Crow, is to
liken it to snow with flakes (lidar points)
‘settling’ on each surface that they contact;
some flakes will be scattered over trees
and bushes and fences, and some will also
reach the ground . If you mentally remove
everything on which the ‘snow’ has settled,
you are left with a cloud of flakes floating
in three-dimensional space (Figs 8 and 9) .
A point cloud is defined by Heritage3D
as ‘a collection of XYZ coordinates in a
common coordinate system that portrays
to the viewer an understanding of the
spatial distribution of a subject’ . The key
thing to remember with regard to a point
cloud is that these are individual points in
Fig 9 Point cloud showing how features can be viewed using enlarged points (lidar © Mendip Hills AONB; source,
Cambridge University ULM (April 2006)).
9
space that have no physical relationship
between them, but because of their density
they can still help to define features .
However, in spite of the fact that the
density of points makes possible a degree
of visualisation, it is not possible to create
shading etc, which makes viewing features
much easier . By contrast, by creating
a surface from the data, they can be
visualised more easily, especially as this
permits the use of lighting effects and
surface analyses such as slope and hillshade generation .
There are two main forms of surface
that can be generated, either creating a
TIN (Triangulated Irregular Network)
direct from the cloud data or a raster
surface indirectly through the creation of
gridded data .
A TIN (Triangulated Irregular Network)
consists of nodes that store the z values,
connected by edges to form continuous,
non-overlapping triangular facets (Figs 10
and 11) . TINs are essentially vector based
and therefore can have a variable area size;
the input features used to create them
remain in the same position as the nodes .
As a result, no extra data are created or lost
through interpolation, so a TIN maintains
all the accuracy of the input data with a
minimum file size, while at the same time
enabling modelling of values between the
known points . Another advantage is that it
is sometimes easier to visualise exactly what
a TIN consists of by looking at a wire frame
image without any surfaces .
A raster surface is different from a TIN
in that it is stored in grid format, ie a grid
of defined cell size is effectively draped over
the point data and each cell is allocated
the z value that falls within it . Because it
consists of a regular array this means that
the points are ‘derived’ from the original
data, rather than comprising the actual
points that were captured in the survey .
Any empty cells have values allocated,
which are derived through the interpolation
of adjacent points . Cells containing
multiple points will be given an average
value . The smaller the cells, the greater
the precision of the grid, or in other words
the higher the resolution of the image .
Fig 10 Wireframe model of the same area showing the
nodes connected by edges (lidar © Mendip Hills AONB;
source, Cambridge University ULM (April 2006)).
Fig 11 TIN surface of the same area showing the effects of
rendering (lidar © Mendip Hills AONB; source, Cambridge
University ULM (April 2006)).
Fig 12 Raster surface of the same area showing the more
natural smoothed surface (lidar © Mendip Hills AONB;
source, Cambridge University ULM (April 2006)).
10
Because values are interpolated into the grid
it is impossible to locate individual features
more precisely than the size of the grid cells .
Care should be taken over the creation of a
raster, as creating cells of a larger size than
the resolution of the data capture will result
in a loss of information .
Equally, while using a cell size smaller
than the resolution of the original data
capture can produce ‘sharper images’, the
interpolation required will create artificial
data in addition to that captured . For
example, if a survey is captured at one laser
hit per metre, creating a grid with a 0 .5m
cell size would result in 75% of the final
data being calculated rather than measured,
and will therefore be less reliable . If two
hits per metre were initially captured, then
a grid of 0 .5m cells would double the
number of data points in the raster . It is
recommended that interpolation does not
exceed this doubling of data .
While it maintains the accuracy of
the original data better than a raster the
TIN is not generally as easy to manipulate
compared to a raster file of comparable
size . In most cases the surfaces produced
by suppliers will tend to be rasters, as they
are simpler to create and fulfil the main
requirements of lidar surfaces (Fig 12) .
Furthermore, many standard GIS packages
require a TIN to be converted into a raster
before any second-order derivatives can be
produced or additional analysis carried out .
The question of TINs will therefore not be
addressed further here .
There are some key differences
between data provided as a point
cloud and data provided as a surface .
As discussed above, and noted in
further detail in Section III 2 .2
below, data can be provided either
as ‘filtered’ or ‘unfiltered’; while
such data can be provided in either
point cloud or surface format they
are much easier to visualise and
understand as a surface . This is
irrespective as to whether the data
are provided as a gridded raster
image or as a TIN . There are pros
and cons to both sets of data that are
discussed in detail below:
Point Cloud
PROS
l
l
All the subtleties are present in point
cloud form; no data have been lost
during the gridding process .
If it is provided as x, y and z data it
can be read by most standard GIS
software .
l
l
With additional three-dimensional
components to GIS or stand-alone
software it is possible to manipulate
the data extensively .
There are no additional processing
costs .
CONS
l
Visualisation and interpretation are
more difficult; requirement to mentally
filter out distractions and imagine how
to join the dots .
Surface
PROS
l
l
l
It is easily readable in standard GIS
software .
Surfaces are much easier to visualise
and make possible hill-shade and
other types of raster analysis .
It facilitates cross-section
investigation of elevated landscapes
and features .
CONS
l
With raster surfaces there is the
risk of some loss of original data
resolution leading to smoothing
away of features or creating a greatly
increased dataset from using smaller
cell size .
Misleading data processing artefacts
may be created .
Depending on the format of
processed data, there will be limited
options for manipulation .
There are additional processing costs .
estimates to be above the natural ground
surface by comparing the relative heights of
adjacent recorded points . DTMs are used
extensively in planning and terrain analysis
for removing buildings etc but really come
into their own in woodland landscapes .
From an archaeological point of view
there is generally little difference between
the DSM and DTM in open landscape,
and interpretation is often easier from a
DSM that has not had buildings and field
boundaries removed, as these can help
in the interpretation and screening out
of features related to modern land use .
Furthermore, the processing of the data to
create the DTM that allows for the canopy
penetration can also smooth out certain
features of archaeological interest (Fig
13) . However, in woodland the DTM is
invaluable . While the last return data from
woodland will penetrate through a degree
of the canopy as compared to the DSM
(Fig 14), as noted above, it will leave a
number of tree trunks etc where the lidar
pulse could not reach the ground surface
and these interfere with visualisation of the
ground surface . By processing this data with
algorithms to create the bare earth DTM,
an unrivalled view of the woodland floor
can be created (Fig 15) .
3.3 File formats
These elevation models can be provided
in a number of formats depending on the
requirements of the end user and on the
l
software that is being used to analyse the
data, so it is important to be clear as to how
the data will be used from the outset .
l
The simplest way to view the data is
as an image, either as hard copy or in a
standard image format as used for digital
3.2 Surfaces DEM, DTM and DSM
photographs (eg Tiff and jpeg) . These are
Of much greater importance is the issue of
usable to a point, but are somewhat limited
the distinction between the types of raster
and do not take advantage of the full
surface available, most specifically among
potential of using lidar data . That said, there
DEMs, DSMs and DTMs .
are situations described below where the
A Digital Elevation Model (DEM) is
use of basic imagery can provide a useful
a form of raster image in which the value
tool for further research and analysis .
assigned to each cell is a height (elevation)
There are some issues dealing
value, rather than a tonal one . This is a
specifically with image files that relate to the
generic term that can refer to both DSMs
and DTMs . In basic terms a Digital Surface different file formats available, such as the
product specific (eg img) versus the generic
Model (DSM) is precisely that; a model
(eg jpeg) . The nature of image files is such
of the surface of the earth (or a section
that they contain different levels of data
thereof) that includes all the features on
often in relation to their file size and the
it such as vegetation, buildings etc . By
questions of format with regard to viewing
contrast a Digital Terrain Model (DTM)
are not specific to lidar applications, and
is a ‘bare-earth’ model . There are various
so are not dealt with in detail here . For
techniques to remove surface features .
further information see http://ec .europa .
Usually, mathematical algorithms are
eu/ipg/standards/image/standard_image_
used to classify the nature of the various
file_formats_en .htm . One other important
returned points into those on the ground
and those off ground . This classification will factor to bear in mind when planning the
use of lidar data and the format in which
aid the removal of all those features that it
l
11
Fig 13 (above) Comparison of lidar DSM and DTM near Alston; note that as well as removing trees there has been a softening of archaeological detail on the DTM particularly
for the lead-mining adit in the foreground and the quarry and small dam in the centre of the image. (© English Heritage).
Fig 14 (top opposite) DSM of an area of woodland in Savernake Forest showing the tree canopy (lidar © Forestry Commission; source, Cambridge University ULM (May 2006)).
Fig 15 (bottom opposite) DTM of the same area clearly showing the course of the Roman road (lidar © Forestry Commission; source, Cambridge University ULM (May 2006)).
12
13
it will be supplied relates to the size of the
files . It is quite possible for a large area
covering several tens of kilometres to be
provided as a single dataset, but the size of
the files may make it impractical to actually
work with . The data file for the survey of
an area 64km2 recorded at 0 .5m resolution
is nearly 1GB in size when provided as an
ASCII grid (256 million cells) . Even on
relatively high-end workstations with highspeed processors and gigabytes of RAM it
is impossible to view the file at its collected
resolution, which reduces its usefulness
considerably . It is preferable to specify that
the data be supplied as discrete blocks
(eg the 2km 2 2km squares as generally
supplied by the Environment Agency)
whatever the format of the data .
Summary
l The data go through many levels of
processing before they reach the end
user; these processes can simplify use
of the data but can also remove useful
information and create misleading data
artefacts . Surface data are generally
much more user friendly and easier
to visualise, but there can be data in
the raw point cloud that are lost in the
processing .
l DSMs, showing landform, buildings
and vegetation, and DTMs, showing a
‘bare-earth’ landform, provide different
information and both have a role to play
in archaeological interpretation .
l In areas of largely open landscape using
the DSM or unprocessed last return data
is preferable to using the DTM .
4 Accuracy and Resolution
As noted above, for archaeologists the key
data recorded with lidar are height data
or more accurately, three-dimensional
coordinates on the ground; it is the height
values that are emphasised because they
make possible the detection of features
of archaeological interest, but the x and
y coordinates are just as important to
accurately locate the features on the
ground . However, it should be noted that
what is actually recorded by the sensor is
only relative data; it is only through the
GPS and IMU recording of the position
of the sensor that it is possible to obtain
absolute coordinates .
There are therefore, two levels of
accuracy that may be given for a given
sensor and/or a given survey: absolute and
relative accuracy . The relative accuracy
of the data are typically in the range
100–150mm, but may be better (often 70–
80mm), and the absolute accuracy depends
on the registration with the datum that is
used . In most cases within England this
will be the Ordnance Survey (OS) grid . In
general, laser-scanned data are registered
initially against WGS84 and it should be
borne in mind that the transformation to
OSGB can create potential distortions
depending on the transformation used .
These issues should normally be
addressed by the supplier and need not be
a cause for concern, but it is worthwhile to
remember that there are potential problems
in absolute accuracy when combining with
other highly accurate data .
It should be noted that in many cases
for the archaeologist it is the relative
difference that is more important than
the absolute, as this difference reveals the
presence of features; it is the fact that there
is an area of ground that is slightly above
or below the surrounding level that reveals
the presence of a bank or ditch . At the first
level of information and interpretation
it is less important whether the feature
is at 120 .25m OD or 122 .25m OD than
whether it accurately depicts the presence
of a previously unrecorded enclosure, but
the difference in absolute accuracy may
lead to difficulties in interpretation and
registration between adjacent lidar datasets
(see Fig 22, showing wavy swathe edge) and
when additional data are recorded using
ground survey techniques with a higher
level of accuracy (eg differential GPS) .
It is not only the accuracy of the lidar
data that needs to be considered, but,
as with any remote sensing technique
applied to the recording and interpretation
of archaeological features, it is also its
resolution . However, unlike imagery where
it is simply the case that any feature that
is smaller than the resolution of the data
will not appear, the issue of resolution with
regard to lidar is more complex because
resolution is a relevant factor at different
stages of the process and is consequently
affected by different specifications .
The resolution of the gridded data
that are used for visualisation is important
because it limits the size of the features
that can be seen and recorded, much in
the same way as for other image-based
data, such as satellite or standard aerial
photography . Figure 16 shows how this
can affect the visualisation of different
archaeological features .
More important for the accuracy of
visualisations, however, is the original
resolution of the data defined by the
number of hits within a square metre and
the footprint (diameter) of the laser beam
when it strikes a surface . As a parameter
largely determined by the flying height of
the aircraft above the ground, footprints
tend to range between 0 .25m and 1m .
If using only a small footprint, say 0 .5m,
and an average of one hit per metre,
measurements will only be taken from 20%
of the ground surface . Because the number
of hits per metre is an average, in a survey
described as ‘one hit per sq m’ it is quite
possible for several squares to have more
than one hit and several to have none .
Fig 16 Effect of resolution on feature visibility – A Roman signal station on Hadrian’s Wall seen at 2m ground resolution (left) and 1m (right) (lidar © English Heritage; source Cambridge
University Unit for Landscape Modelling (March 2004)).
14
Fig 18 Point density: gridded versus as captured (lidar © Mendip Hills AONB; source, Cambridge University ULM (April 2006)).
data were being used to monitor the
erosion of a promontory fort (Fig 21) .
Here comparison with ground-based
GPS survey showed some substantial
discrepancies in the position of the cliff
edge, which are most easily explained by
the assumption that certain pulses were on
the extreme edge of a given 2m square and
missed the cliff entirely .
Careful planning at the data capture
stage can minimise later difficulties with
resolution . In wooded landscapes Forest
Research and Cambridge University’s
Unit for Landscape Modelling have been
doing the surveys at two hits per metre
and gridding to four, ie 0 .5m ground
resolution . They also have a fairly large
footprint to maximize the chance of getting
a reflection from the forest floor and
transects have a 65% overlap to ensure
good coverage .
Note that all the airborne lidar data
discussed thus far has related to data
collected from fixed-wing aircraft, which
due to restrictions on speed and altitude
are limited to collecting data in the range
of up to eight points per square metre with
each pass; it is possible to collect more
points by carrying out multiple passes,
but this has implications for flight time
and consequently for costs . There are also
systems on the market that are designed to
be mounted on helicopters that can collect
much higher densities of data . The Fli-map
and ATLAS systems were deployed by
DEFRA (Department for Environment,
Food and Rural Affairs) for measuring
changes in beach levels and recorded
between 12–28 points per metre (McCue et
al 2004), while the Discovery Programme
in Ireland surveyed Dún Ailinne prehistoric
hillfort in Co . Kildare at 15–30 points per
metre and the Hill of Tara at 60 points
per metre using Fli-map 400 (Corns et
al 2008) . This higher resolution shows a
much greater degree of detail, but comes
at the price of generally smaller areas being
flown .
the bluestones in the north-east quadrant .
Figure 20 shows the gridded data
against a rectified aerial photograph .
The outlines of the stones are coloured
according to the number of points that fall
within their outline; those in green have no
strikes at all . Although the point data used
here is gridded rather than the original
point cloud it demonstrates the same point:
that small features can be missed or equally
small features (eg sheep) can affect the
apparent height of the ground surface!
A second example of this effect is
found on the Welsh coast where lidar
Summary
l From an archaeological point of view
relative accuracy is often more important
than absolute accuracy .
l The relative accuracy of the data is
typically in the millimetre range (100–
150mm), but can be higher .
l The absolute accuracy of the recorded
data is heavily dependant on the
accuracy of the GPS .
l The resolution of the gridded data is
important because it limits the size
of the features that can be seen and
recorded; 2m resolution data are
Fig 17 Point density, showing the actual distribution of points over an area (lidar © Mendip Hills AONB; source,
Cambridge University ULM (April 2006)).
Figure 17 shows the actual distribution of
points over an area of field, while Figure
18 shows the difference between the actual
points as captured (red) and the resulting
grid (blue) .
A good example of this is shown by the
stones at Stonehenge . The lidar imagery
captured at one hit per square metre
does not appear to show several of the
bluestones . Figure 19 shows the outline of
the main stones within the henge in pink
against the surface model generated from
the lidar data captured at one point per
metre . There is no clear trace of several of
15
Fig 19 Stonehenge bluestones: comparison of the lidar data with the known stone positions shows several are missing (lidar © Environment Agency
(December 2001)).
Fig 20 Stonehenge bluestones: gridded lidar data versus the actual position of the stones (lidar © Environment Agency (December 2001); aerial
photograph © English Heritage. NMR NMR24182/003 (01-MAR-2006)).
16
Fig 21 Linney Head promontory fort, Pembrokeshire showing the discrepancy between the lidar modelled data and the
ground based GPS survey. (NPRN 94226, © Environment Agency copyright, D0055624. All rights reserved. View generated
by RCAHMW).
l
l
l
generally inadequate for recording many
archaeological features; 1m resolution is
the basic minimum, but where greater
detail is required higher resolution data
are preferable .
Survey in woodland requires higher
resolution data (typically two hits
per metre gridded to 0 .5m ground
resolution) to achieve sufficient canopy
penetration .
Original point density is as important as
final resolution, as insufficiently densely
spaced points can risk missing features
altogether .
Low-level survey can record points up
to a density of 60 points per metre, but
for large area survey one or two points
per metre (gridded to 1m resolution)
is adequate to record most features of
interest .
Part II How to decide if you need
it: practicalities and limitations
1 Project planning
(see decision tree on page 39)
1.1 MoRPHE
The potential for lidar data to contribute
to a project should be identified as early as
possible . Under Management of Research
Projects in the Historic Environment
(MoRPHE) guidelines (Lee 2006), its use
should be assessed as part of the project
design document . Advice may be required as
to whether the site or landscape in question
is appropriate for the use of lidar survey
and whether it will yield useful results . In
England this advice can be sought in the
first instance from the English Heritage
Aerial Survey and Investigation (AerSI)
team or from the relevant English Heritage
Regional Science Advisor . More technical
advice may also be obtained from the Aerial
Survey and Investigation Team, and the
Archaeological Survey and Investigation
team could advise on the likely cost benefits
of alternative terrestrial survey techniques
and archaeological interpretation, especially
if the survey area is quite small or if the
level of detail required is higher than will be
readily achievable using lidar data . If your
survey area covers a largely wooded area,
then technical advice may also be obtained
from Forest Research, some of whose staff
have a particular expertise in this area .
1.2 Survey considerations and options
As with any project, one key element before
any work is undertaken is to be clear about
the objectives, requirements and end-use
of any lidar data . While lidar as a technique
has been around for some time its use is
still relatively new for archaeologists and
while it is particularly useful in certain
situations and can produce spectacular
results (Bewley et al 2005 and Devereux et
al 2005) it is less useful in other situations
and always needs careful interpretation
(Crutchley 2006) .
A key point to remember is that lidar
primarily records height information,
therefore the features being surveyed
must have a three-dimensional surface
aspect . As noted above, the intensity data
from the lidar return are able to record
certain aspects of the reflective nature
of the surface recorded, which may
provide information on factors such as
angle, roughness, dampness and colour
absorbency, but only in exceptional
cases will this information directly reveal
archaeological features .
The bottom line, however, is that lidar
does not penetrate the ground . If the
archaeological features of interest are not
represented on the ground surface then
lidar will not be able to record anything
except the general topography of the survey
area . Of course, having an accurate record
of the general topography of an area and
the surrounding landscape can be a useful
resource in itself, but if this is all that is
required then lidar may not be the most
appropriate, or cost-effective, method with
which to collect these data .
Basic topographic height data at scales
suitable for general topographic relief
are available from alternative sources .
Depending on the resolution required there
are commercial datasets available, such as
those from the Ordnance Survey (http://
www .ordnancesurvey .co .uk/oswebsite/) or
from NextMap (http://www .xyzmaps .com/
NextmapTerrainData .htm), or even freely
available from the web (eg US Geological
Service http://seamless .usgs .gov/ or NASA
https://wist .echo .nasa .gov) .
Once it is clear that there are likely to
be features that can usefully be recorded
by lidar, the next stage is to be clear
about the end use of the data . Is the lidar
data required as the primary source,
an interrogatable dataset that can be
analysed by different staff to provide an
interpretation of archaeological features,
or is it seen as a background layer for
other datasets available elsewhere? This
decision will determine the form in
which the data will be provided, which
will in turn dictate the requirements for
software and hardware .
The precise nature of these options is
discussed in more detail below .
If the aim is simply to use the surface
model derived from the lidar data as
a background layer, the hardware and
software requirements are probably quite
low, but the processing of the data to an
appropriate format for GIS etc (see below)
will need to be budgeted for . If, however,
the intention is to analyse the data in-house
and carry out any type of interpretation,
then the appropriate hardware and
software must be available to deal with the
large datasets . While it is possible to view
the processed data that are provided by
most suppliers in standard GIS packages
such as MapInfo or ArcGIS, without
17
specialist extension modules this is not a
very user-friendly option (see below) .
This is not just a question of hardware
and software, but also of technical
expertise; familiarity with the process of
generating and manipulating digital models
is necessary . Similarly, the interpretation of
archaeological features from these models
is best done by someone experienced in
the interpretation of aerial data, especially
if the intention is to look at other sources
at the same time, something that is
recommended for reasons given below .
If there is the further requirement
to actually map archaeological data (or
indeed any other type of feature) from the
lidar data, then this presents a different
set of problems . Until recently there were
no simple tools for mapping directly from
processed lidar data, ie derived surface
models that can be manipulated to control
height exaggeration and lighting position
(see Interpretation) and English Heritage
had to develop their own flowlines and
working practices . However, given that
software and hardware capabilities are
changing all the time it is wise to consult
someone already actively working with
such processed data .
Summary
l Advice on whether lidar can be useful
for a given landscape can be obtained
from the English Heritage Aerial Survey
and Investigation team or from the
relevant English Heritage Regional
Science Advisor .
l More technical advice on the use of
lidar data may also be obtained from the
Aerial Survey and Investigation Team .
l The English Heritage Archaeological
Survey and Investigation team can
advise on the likely cost benefits of
alternative terrestrial survey techniques .
l If the survey area covers a largely
wooded area, then technical advice
can be obtained from Forest Research,
some of whose staff have a particular
expertise in this area .
l Basic topographic height data at scales
suitable for general topographic relief
are available from alternative sources,
for example the Ordnance Survey or
NASA .
l It is important to be clear as to whether
the lidar data are required as the
primary source or whether it is seen as
a background layer for other datasets
available elsewhere .
l To make best use of lidar for
archaeological survey the project team
should include someone with suitable
experience in using aerial data .
18
2 Where can you use it?
2.2 Moorland
One of the major factors affecting the
usefulness of lidar is the current landuse of the area of interest, as this can
have a major impact on the survival and
consequent visibility of features .
Moorland is another landscape type
where ground survey is often difficult and
dependent on the season . Typical moorland
vegetation such as bracken and heather
can make the surveying of features on the
ground difficult and limit the window of
recording to certain times of the year . The
timing of any lidar survey flight is likely to
be of particular importance . Although it
is possible that the use of last-pulse data
will enhance the visibility of features under
heather and gorse during autumn and
winter it is likely that at other times they will
prove too dense for the beam to penetrate .
It has been suggested that there may
be issues with regard to the use of lidar on
open-stone landscapes or on features created
from stones, for example cairns, rock waste
mounds . So far English Heritage staff have
limited experience in such landscapes, but
a project currently underway in the North
Pennines may help clarify matters .
2.1 Grassland
Many archaeological earthworks are found
in areas of open grassland and lidar can be
a useful tool in such landscapes . Although
archaeological aerial reconnaissance and
field survey have often targeted such areas
in the past to great effect, and continue to
do so, the manipulability of lidar data can
prove a valuable additional tool . This is
particularly the case for improved pasture,
one of the more difficult types of landscape
for survey by other means because the
ploughing has eradicated most traces of
any former earthworks, but the presence of
grass as opposed to an arable crop restricts
the possibility of cropmarks to periods of
extreme drought . However, if there are any
traces of earthworks surviving, even in a
smoothed and eroded state, then lidar is an
excellent tool for recording them . This is
true for all forms of grassland ranging from
upland grass and stone landscapes such as
the Yorkshire Dales, to coastal saltmarsh .
2.3 Arable
Landscapes currently under arable
cultivation are generally the most
responsive when it comes to conventional
aerial photography and survey . Given
the right conditions they will produce
Fig 22 Lidar showing palaeochannels in the Witham Valley. Note also the two wavy lines running down the image; these
are processing artefacts resulting from the overlapping of adjacent data swathes. (lidar courtesy of Lincolnshire County
Council; source, Environment Agency (March 2001)).
evidence of former activities in the form
of cropmarks and soilmarks . By contrast,
they are probably the worst for analytical
field survey because any earthworks have
been consistently eroded by the plough,
until there are very few surface traces left .
Lidar can still recover some information
from such landscapes because even where
former banks and other features have
been heavily eroded and are only visible
as broadly spread features raised about
100mm above the surrounding ground
level they still have a surface expression .
The capacity for lidar to look at large
areas and pick out patterns, together
with the ability to manipulate the data to
enhance slight features, means that it is
possible to record features that would be
almost impossible to locate on the ground
in a ploughed field . However, while lidar
may be successful in showing extensive
features such as field systems it is likely to
be less successful for discrete features such
as barrows or enclosures .
The majority of cropmark sites are
unlikely to have any other significant
surface expression of the buried features
and so lidar height data will not be able to
identify them . Although, as noted above,
there may be some potential for lidar
intensity data to reveal cropmarks, and
there is a chance that if the cropmark itself
has sufficient height difference this might
register in the lidar first return data .
An understanding of surface geology
is important, as in many arable areas,
particularly those where there has been
significant deposition such as flood
plains, the results will be less successful,
something that is equally true for
traditional aerial photography . On the plus
side, the majority of low-lying arable areas
will already have some lidar data through
the Environment Agency’s policy of
recording river valleys, so these data might
be available more cheaply than having to
commission a new survey .
The archaeological value of lidar in
revealing geomorphological features (Fig 22)
should not be underestimated, particularly
in the main river valleys and in the fenlands,
where much of the Environment Agency
survey work has been targeted (Challis 2006
and Jones et al 2007) .
2.4 Woodland
The key area of land use where lidar
comes into its own and has substantial
advantages over other forms of survey is
woodland . The efficacy of the technique
is demonstrated in the case study on
Savernake and in the separate section on
woodland below Part V .
3 To map or not to map?
As noted above, one of the key questions
with regard to the use of lidar data is
whether it is planned to actually map from
the data, or just to use it as a background .
For specific site surveys a case can be
made for using lidar derived imagery as the
basis of a field survey . This enhances and
improves planning of site details in the field
based on of a combination of field survey
and image interpretation . Alternatively
the lidar derived imagery can provide a
useful topographical background against
which survey can be carried out . This is
particularly the case in areas of ancient
rivers where lidar provides an excellent
source for palaeoenvironmental data that
can in turn aid in the interpretation of
sites based on their location (see Part IV 2,
Witham Valley case study) .
It should also be borne in mind that
the data from the lidar survey can be
useful apart from simply interpreting
readily visible archaeological features .
For example, because of the high level
of detail provided by the data, it can be
used to compare the relationship of the
man-made rampart slopes to the steepness
of the underlying topography in upland
areas . It can be used to assess the local
topography and how this might have
affected movement or supply, such as
confirming the practicality of a given route
for an aqueduct . It should, however, be
noted (see Part II 1 .2) that there are other
sources of data available that provide basic
topographic data at a range of resolutions .
These can be derived from other sensors,
such as radar (see NextMap) or by using
photogrammetry from conventional aerial
photographs (eg as part of the Cawthorn
Camps survey, Stone 2004) .
However, in most cases the extensive
dataset provided by lidar is probably
best treated as one of the sources for a
desktop survey, maximising the value of
the initial cost of the product and making
it possible to target more expensive
fieldwork more carefully . Interpreting the
lidar data into a mapped form ensures
that the data are fully examined and that
the archaeological results are properly
documented . The quality of interpretation
and metrical accuracy possible from lidar,
used in conjunction with air photos and
other sources, provides a high degree of
confidence in the results . Field survey can
then be used to examine those areas where
there is a lower degree of confidence,
for example stratigraphic relationships
between features, areas with poor visibility
in available datasets or confused by surface
features such as dense undergrowth or
piles of forest residue, or areas where the
complexity of remains and management
issues can only be addressed through
direct observation .
The results from the Savernake Forest
survey (see Savernake case study) suggest
that for continuously wooded areas,
using lidar is likely to be the best single
remote sensing method and therefore a
combination of lidar and field checking
may be an appropriate methodology .
However, if there are areas of open
ground within the area of survey (or
there have been in the last 50 years),
then it is likely that checking the available
air photo sources will be of significant
benefit . Generally this is best done in
parallel with the lidar analysis, but for
some projects a staged approach may be
more appropriate .
In areas of mixed woodland and arable
the air photo sources will be essential to
ensure the best possible understanding
of the archaeology . A survey comparing
existing HER information and a selection
of aerial photographs with lidar data was
carried out by Birmingham University and
concluded that both sources are required
for a complete picture of the archaeological
remains of any given area (Challis et al
2008b) .
Summary
l In many cases the extensive dataset
provided by lidar is best treated as one
of the sources for a desktop survey to
produce an interpretative map of the
features identified .
l The quality of interpretation and
metrical accuracy possible from lidar
(used in conjunction with air photos and
other sources) provides a high degree of
confidence in the results and makes it
possible to target fieldwork carefully .
4 Data acquisition
One of the first steps when planning to
acquire lidar data is to assess whether the
data for your area of interest already exist .
As noted above, the Environment Agency
has been carrying out lidar surveys around
the coasts and river valleys for about
ten years and these data are available to
purchase from them either as hill-shaded
jpeg images or as the actual gridded data .
A catalogue of their holdings – updated
regularly – is held by the Environment
Agency and available for download from
their website (http://www .geomatics-group .
co .uk/GeoCMS/Order .aspx) .
Other commercial companies have also
been carrying out lidar surveys for several
years throughout the country, as have
19
research companies such as the Unit for
Landscape Modelling at Cambridge http://
www .uflm .cam .ac .uk/ and the Natural
Environment Research Council (NERC)
http://www .nerc .ac .uk/ .
Unfortunately, because the majority
of these projects were carried out on
behalf of paying clients the data are not
readily available and indeed there is
no central record of where the surveys
cover . However, there are exceptions
and organisations such as the Forestry
Commission are in the process of collating
a central record of the lidar data that they
hold . The general lack of coordination
is an issue that is being considered by
Heritage3D and the Remote Sensing and
Photogrammetric Society (RSPSoc) . There
are some international sites on the web
that claim to have records of general lidar
cover such as http://www .lidardata .com .
These are currently very much restricted
to the United States, but it is possible that
a commercial company may fill this gap in
the UK in the not too distant future .
If no data exist for your area of
interest, or if the data that do exist are of
insufficient quality for any reason (it may
be of insufficient resolution or simply
too old etc), then it will be necessary to
commission a new survey . It is worth
bearing in mind, that a large number of
lidar surveys are carried out each year for
non-archaeological purposes, for example
for infrastructure planning . For many
large infrastructure projects such as roads
or pipelines that are covered by PPG16
(or its successor) it is quite possible that
a lidar survey may be commissioned by
the developer to establish the nature
of the topography of the area and for
other reasons . Surveys may even be
commissioned by local authorities or other
bodies that have links with archaeological
organisations, as lidar can have a significant
role to play when first appraisals of large
landscape developments are undertaken,
say for EIA (Environmental Impact
Assessment) planning stages .
The level of detail required for such
surveys would most probably be sufficient
for archaeological needs (especially for
open landscapes), but it is worthwhile
trying to influence any project you are
aware of so that it provides the most useful
data . This can be particularly important in
either wooded or moorland areas where a
survey carried out in the height of summer
when all vegetation is at its densest
will be less useful than one targeted for
winter or spring when vegetation is less
prevalent . Equally, if a heritage-based lidar
survey is being considered, there may be
20
other potential users of the data in other
disciplines or organisations and it may be
possible to collaborate on a project .
A lot of the elements with regard to
commissioning any type of laser scanning
survey were addressed by the Heritage3D
project, and the guidance document
resulting from it includes a section on
the commissioning of an airborne lidar
survey, finding a contractor and ensuring
that the survey is carried out to the correct
standards (‘Developing professional
guidance: laser scanning in archaeology
and architecture’, English Heritage 2007
http://www .heritage3d .org/) .
What is not currently specified in that
guidance is what those standards and
specifications should be when lidar survey
is to be used for examining archaeological
sites and landscapes . As noted, one of the
key factors is the resolution of the data
defined by the point density on the ground .
The point density is defined by Heritage3D
as the average distance between the x, y
and z coordinates in a point cloud, and
for lidar this refers to the number of hits
on a surface within a one-metre square for
the raw data . While it is fairly evident that
a higher ground resolution is likely to be
able to record more features, the cost of
obtaining and using these larger datasets
also needs to be borne in mind .
Some of the variables that determine
the resolution of the data are defined
by the aircraft, such as altitude and
ground speed; others by the lidar system,
including laser frequency (pulses per
second), scan frequency and scan angle .
The actual laser frequency of the system is
generally fixed, but it has been increasing
over time and is likely to get faster . Early
lidar systems had a frequency of only 10–
15KHz, whereas today there are systems
capable of up to 250KHz (ie 250,000
points recorded every second) .
With a fixed scan frequency, in
order to increase the point spacing it is
necessary to reduce the altitude of the
aircraft, which is often impossible because
of aviation regulations; to fly with larger
overlaps, which increases flying time and
hence costs; or to reduce the scan angle .
Reducing the scan angle reduces the swath
and increases the number of passes that
need to be flown, again increasing the cost .
At an altitude of 1000m a 15° scan angle
produces a swath of 536m; a scan angle
of 7° produces a swath of only 246m . The
Unit for Landscape Modelling (ULM) at
Cambridge University provides a useful
calculator on their web site (http://www .
uflm .cam .ac .uk/lidar .htm) to assist in
planning surveys
Summary
l Step one – assess whether the data for
your area of interest already exist – see
Environment Agency catalogue etc .
l A large number of lidar surveys
are carried out each year for nonarchaeological purposes, such as for
infrastructure planning .
l Lidar can have a significant role to
play when first appraisals of large
landscape developments are being
undertaken, say for Environmental
Impact Assessment (EIA) planning
stages .
l A lot of the elements with regard to
com-missioning a laser scanning
survey were addressed by the
Heritage3D project .
l The Unit for Landscape Modelling
(ULM) at Cambridge University
provides a useful calculator on their
web site to assist in planning surveys
by estimating the total flight time
required .
l Large, rectangular survey areas are the
most cost effective, having the minimum
number of turns at the end of each
aircraft run; small or irregularly shaped
areas are less cost effective .
l Ensure that you know the actual form
in which the data will be provided; it
is no good obtaining data from a
contractor if it is in a format that the
end user cannot use .
5 Dissemination, archiving and copyright
It is essential that all issues relating to
dissemination, archiving and copyright
are considered at the outset of a project .
This will ensure clarity in what data and
imagery it is possible to publish or make
available to others for future research .
5.1 Dissemination
Lidar data files and the generated
imagery are generally quite large files;
the gridded ASCII text files for 2km 2
2km tiles at 1m resolution are in the region
of 150Mb, and the original ungridded
data c 250Mb . As such they are not easily
supplied to third parties or colleagues .
They can also prove taxing to process for
lower specification PCs .
It must also be remembered that the
actual lidar data are often copyright to
a third party and so cannot be distributed;
however, while the actual data are usually
strictly controlled, this is not always the
case with the imagery generated from them .
Another key point to remember is
that in terms of useful data from a given
survey, the key product is the interpreted
layers and attached records . In most
cases, however, it would be good practice
to support this with at least a layer of
uninterpreted information, for example a
hill-shaded image etc (where available) .
5.2 Archiving
Heritage3D raised many questions about
appropriate formats for long-term storage
of data . As this is an area that is constantly
developing, advice should be sought at the
outset of a project from ADS (see Bewley
et al 1998; or http://ads .ahds .ac .uk/project/
bigdata/) . If the copyright of the data is
held by a third party then the question
of what derived products – such as hillshaded images – can be archived should
also be addressed at an early stage .
Part III – How do you use it?
There is no single answer to the question
of how to use lidar data . Much will be
determined by the nature of the survey
and by the technical equipment, or lack
thereof, available to those doing it . For
some surveyors it may be appropriate to
rely solely on a hard copy of a hill-shaded
image provided by others; for others
such an image may be viewed using a
portable GPS device . Elsewhere, where the
hardware and software are available, onscreen desktop analysis is preferable, as this
makes it possible to view and manipulate
the data to maximise its interpretational
value . In all cases the surveyors need to
understand how to interpret the visual
evidence and be aware of likely pitfalls .
5.3 Copyright
Several questions have been raised about
the nature of copyright with regard to
lidar data . The most important of these
is to what extent does the ‘added value’
of creating hill-shades etc create copyright
in the hands of the author of those
images? Unfortunately this issue is still
not entirely clear .
When commissioning work, however,
or when getting data off the shelf, it is
important to try to be clear from the
outset where copyright lies and how the
data can be disseminated . Copyright of
any images generally resides with whoever
created them, unless a different specific
arrangement was made . But even in
this situation the data source is usually
acknowledged as a courtesy .
There are inevitable copyright
restrictions on lidar data and costs of
purchasing existing data can vary according
to the size of the area in question, the
resolution and its age . However, for
most archaeological purposes, the use of
older data may not be an issue . Equally,
if the primary archaeological requirement
is to examine features from hill-shaded
images, it is often possible to acquire
these at significantly lower costs than
the fully manipulable elevation data; and
copyright on any images may also be
more relaxed .
Summary
l It is essential that all issues relating to
dissemination, archiving and copyright
are considered at the outset of a project
to ensure clarity in what data and
imagery it is possible to publish, make
available to others .
l Lidar data files and the generated
imagery are generally quite large files
and as such they are not easily supplied
to third parties .
1 Visualisation
Probably the key aspect that determines
the usefulness of lidar data and how they
are used in relation to archaeology is the
question of how the data are viewed . If
the user does not have the facilities to
view and manipulate the original data in a
specialist package, it is still possible to use
two-dimensional snapshots of the data as
standard jpeg or Tiff files . As noted above,
these are somewhat limited compared to
being able to view and manipulate the
data, but they can still be useful . It should
specifically be noted that what may at first
appear to be a relatively unpromising image
in greyscale, or even in colour (Fig 23), can
reveal a considerable amount of ‘hidden’
information after some basic enhancement
techniques – such as equalisation, available
in standard image processing packages –
have been applied (Fig 24) .
Assuming that there is access to
software and hardware to view the data
and that hill-shaded images alone have not
been specifically requested, the standard
product from a lidar survey is likely to
be an ASCII grid . This can be read in a
standard GIS (Fig 25), but is not very
user friendly; it requires specialist threedimensional viewing modules (see Glossary
for a definition of three-dimensional in
this context) or separate programs to really
enable interpretation .
There are a number of threedimensional viewing programs on the
market, ranging from freeware available
over the web (such as Landserf http://www .
soi .city .ac .uk/~jwo/landserf/) to specialist
corporate viewing and modelling software
(such as Applied Imagery’s Quick Terrain
Modeler http://www .appliedimagery .com/,
Terrascan http://www .3dlasermapping .
com/uk/airborne/software/terrascan .htm
or the three-dimensional Analyst module
21
Fig 23 Standard jpeg lidar image from the Environment Agency (© Environment Agency copyright 2008. All rights reserved).
Fig 24 The same image after equalisation; note the number of additional features visible, particularly the short stretches of
bank in the pale green fields in the centre left of the image. (© Environment Agency copyright 2008. All rights reserved).
22
in ArcGIS) . It is not practical to give a
complete listing of all available software
as this is constantly changing, but a quick
search on the internet will list those
available .
(Note: The mention of any specific
hardware or software used by English
Heritage does not necessarily imply
endorsement of this product over any
other, but simply reflects the current
limited use by staff within English
Heritage .)
Without the use of some form
of viewing software, either to view
interactively or to create hill-shaded models
(Fig 26) etc, the point cloud data, DEMs
or DTMs are not particularly useful . With
the software it is possible to view and
manipulate the data and generate your own
images, highlighting specific features (and
even view them truly three-dimensionally) .
Ideally the three-dimensional data
should be viewed stereoscopically, taking
advantage of the brain’s natural ability to
interpret three-dimensional objects aided
by the opportunity to stretch and light
the surfaces differently . This can be done
using specialist viewing packages and
photogrammetric packages using a variety
of special glasses (anaglyph, polarized,
flicker) – although it is to be noted
that special screens are now becoming
available that do not require these glasses .
However, for most users this is another
level of expense that is not feasible, and
the viewing of flat two-dimensional
images on paper or on-screen – or better,
as interactive three-dimensional images
on screen – is the easiest way to view the
data; although described here as ‘three
dimensional’ these are actually ‘two-anda-half-dimensional’ – a two-dimensional
representation of three-dimensional data
(see Glossary) . If the interpreter can be
provided with a flexible tool for viewing,
manipulating and mapping from the
gridded data in real time then many of the
limitations of two-dimensional data can be
addressed .
Such software can be used to visualise
different lidar-derived datasets in a
number of ways (see Data Formats [above]
and Algorithms [below]) . The most
obviously user-friendly product, because
it is relatively easy to interpret, is the
hill-shaded image . One of the key benefits
of the surfaces derived from lidar data
is the fact that like all DEMs it is possible
to manipulate them with various software
packages to produce images lit from
any conceivable position, even from
positions impossible in nature . Coupled
with the ability to increase the vertical
exaggeration of features it is possible to
visualise features that have only a very
slight surface indication on the ground .
Nevertheless the very possibility of
viewing features from a multitude of angles
and lit from a variety of positions can cause
complications . Those used to viewing aerial
photographs are used to the fact that when
features on the ground run parallel to the
direction of the light source they do not
create a shadow and are therefore virtually
impossible to view . This situation is one
that also occurs with lidar-derived imagery,
as is demonstrated in Figures 27 and 28,
which show two blocks of medieval ridge
and furrow cultivation .
The obvious way around this problem
is to produce multiple images lit from
different directions . However, if you are
dealing with hard copy paper images
and looking at a large area, this practice
soon becomes impractical . Some success
has been achieved in creating composite
images using the transparency tools
within image editing or GIS packages
(see Figs 51, 55 and 56), but a more
effective process is now seen in the use
of principal component analysis (PCA)
a statistical method to examine multiple
hill-shaded images and compile a
composite image that shows the main
features from each image (see Fig 54)
(Devereux et al 2008) . There are two
issues with this procedure, however:
first, while there are a number of off-theshelf packages that will create PCAs,
they do not necessarily produce the best
results, which can really be obtained only
by those with a degree of experience
of working with the process; secondly,
whereas a single hill-shaded image is
relatively easy to interpret for anyone
used to viewing aerial photographs,
PCA images with their multiple colours
and sometimes conflicting shadow and
Fig 25 Standard greyscale raster image in Arc (lidar © Mendip Hills AONB; source, Cambridge University ULM (April 2006)).
Fig 26 Hillshaded image in Arc (lidar © Mendip Hills AONB; source, Cambridge University ULM (April 2006)).
Fig 27 Ridge and furrow near Alchester illuminated N–S (lidar © Cambridge University
ULM (Dec 2005)).
Fig 28 Ridge and furrow near Alchester illuminated E–W (lidar © Cambridge University
ULM (Dec 2005)).
23
highlight patterns can be more misleading,
particularly when it comes to interpreting
whether a feature is positive or negative .
This is very important when it comes
to interpreting the function of a feature,
as there is a major difference between a
mound and a hollow . However, altering
the colours of PCA images or draping a
semi-transparent standard hill-shaded
image over the top can help to compensate
for this potential for misinterpretation .
There have also been attempts to
improve hill-shaded models in other
disciplines by combining models lit from
different directions and then giving
weight to the different slopes . While these
may have some benefit for other disciplines,
because the features of interest to
archaeologists tend to be occur irrespective
of the direction of slope this methodology
seems less valuable in archaeological
applications . It also requires a greater
degree of knowledge of sampling and
weighting techniques common to GIS
than most archaeologists have (Loisios
et al 2007) .
In his assessment of the use of hillshaded images in the field for the rapid
recording of features in woodland,
Hoyle 2008 states that they ‘enable the
extent and location of recognised features
to be simply recorded with reference
to the visible features, generally by direct
tracing, and no further surveying is
necessary .’ He adds that ‘this not only
improves the accuracy of the recording
but also significantly speeds up the time
needed to locate, survey and record
identified features, and its cost benefit
cannot be overstated . The hill-shaded
images also present an accurate and up
to date map view of the ground surface,
which is often more comprehensive
than the mapping available from the
Ordnance Survey, particularly of areas
of woodland .’
While hill-shaded images and PCA
images are the principal easily interpreted
forms there is also some promise from
slope models (Doneus and Briese 2006) .
However, as with PCA imagery these
models are not as simple to interpret,
as the natural reaction is to see them
as hill-shades, which can again lead to
interpretational mistakes regarding positive
and negative features . This is clearly an
area with promise, as it is the slope of
a feature, or rather the change in slope
from what appears to be the underlying
topography, that alerts one to the presence
of man-made features, but further work is
required to demonstrate the full potential
of the technique .
24
Summary
l The way lidar data are used will be
determined by the nature of the survey
and by the technical equipment available
to those doing it .
l The standard digital product from a
lidar survey is likely to be an ASCII grid;
this can be read in a standard GIS, but
requires specialist three-dimensional
viewing modules to really enable
interpretation .
l There are a number of three-dimensional
viewing programs on the market ranging
from freeware available over the web
to specialist corporate viewing and
modelling software .
l Ideally the three-dimensional data
should be viewed stereoscopically, taking
advantage of the brain’s natural ability to
interpret three-dimensional objects .
l For most users the viewing of flat twodimensional images in paper or onscreen – or better, as interactive threedimensional images on screen – is the
easiest way to view the data; although
described here as ‘three dimensional’
these are actually ‘two-and-a-halfdimensional’ – a two-dimensional
representation of three-dimensional data
l The most obviously user-friendly
product is the hill-shaded image that can
be produced as lit from any conceivable
position, even from positions impossible
in nature .
l PCA makes it possible to create a single
composite image that shows the main
features from each different lighting
angle .
l While hill-shaded images are the
principal easily interpreted forms there is
also some promise from slope models .
2 Interpretation
2.1 Archaeological
Like an aerial photograph a lidar-derived
image often appears misleadingly simple
to interpret; however, to ensure the best
results from a survey the interpretation
must be done by someone with the
necessary skills and experience . This
becomes even more important if there is
an intention to do any more than mark
‘dots on maps’ and if the work is not
going to be followed up with total ground
survey . There will usually be a significant
cost benefit in detailed evaluation of the
imagery prior to any field survey work .
Lidar data and imagery viewed as
hill-shaded images appear similar to
vertical photographs of earthworks lit by
low sunlight, so the analysis of lidar for
the identification and characterisation of
archaeological sites requires similar skills
as those applied to air photo interpretation,
for example the ability to recognise slight
earthwork banks or ditches based on their
appearance with reference to shadows and
highlights, while filtering out features due
to modern agricultural practices, geology
and data processing artefacts . However,
the lack of any colour or of tonal variations
due to the type of vegetation and other
surface cover can either aid interpretation
or make it more difficult, depending on the
particular feature involved .
As noted above, the basic data recorded
by lidar is height data, and as such there
are no colour data to aid interpretation .
To those unused to interpreting such data,
the wear pattern around an animal feeding
station may look like a small barrow or a
sewage works, defined by low banks, will
appear little different to a small enclosure
(Figs 29 and 30) . Without the use of other
sources it is very easy to make erroneous
judgements, even for those used to dealing
with aerial photographs . During an early
project using lidar data, failure to examine
all available sources at the outset almost
led to a major misidentification of a site
(Crutchley 2006) .
Experience in interpreting aerial
imagery will help to ensure that the sorts
of features caused by either geological
activity or by recent farming practices can
be filtered out, and ensure that the effects
of different lighting angles are used to
best effect to reveal subtle features . For
predominantly non-wooded landscapes,
the possibility of commissioning a mapping
survey using sources other than just lidar
– for example a full aerial photographic
survey using both historic and modern
photographs – should be considered, as
often the interpretation process is made
much easier by comparing different sources .
Aerial photography in England will
normally only be able to photograph
earthworks lit by the sun from the west,
south or east, and requires the photographer
to be there at the right time to make
the record . The great advantage of lidar
data is that they make it possible to view
archaeological earthworks with the light
coming from any direction or elevation .
This gives much greater confidence
in interpretation and can often reveal
previously unseen features . Because the
user will normally be mapping from a
two-dimensional image, it is essential to
remember in which direction the light is
falling to enable the difference between cut
and raised features to be correctly identified .
(If the data are being viewed stereoscopically
this is less of a problem) . Most people find it
easiest to interpret an image when the light
Fig 29 Feature misinterpretation: lidar derived image (lidar © Forestry Commission;
source, Cambridge University ULM (March 2004)).
Fig 30 Feature misinterpretation: aerial photograph showing the true nature of the feature
(© Cambridge University ULM. zknpp0001 02-FEB-2005).
Fig 31 Feature misinterpretation: lidar derived image (lidar © Mendip Hills AONB; source,
Cambridge University ULM (April 2006)).
Fig 32 Feature misinterpretation: aerial photograph showing the true nature of the feature
(photo PGA_ST5050_2006-04-30_part. Licensed to English Heritage for PGA, through
Next Perspectives™; OS background map © Crown Copyright. All rights reserved. English
Heritage 100019088. 2009).
falls from the top of the image as viewed and
the tradition in hill-shading for maps is with
an imaginary sun in the north-west .
Figure 31 shows some of the difficulties
of interpretation from a single hillshaded image . In the top right corner of
the image are number of features with
highlights to the south-west (north is to
the top of the image) and shadows to the
north-east . By contrast, in the bottom
centre is a feature with highlights to the
north-east and shadows to the south-west .
Without reference to other information,
or knowledge of the direction of lighting
it is not immediately apparent which are
negative and which are positive features .
Once the correct three-dimensional aspect
of the features has been acquired, the
feature in the bottom centre gives every
appearance of being a burial mound,
being of a similar size and shape to other
known barrows in the vicinity . However,
the evidence from aerial photographs and
mapping (Fig 32) reveals that this is in fact
the site of a covered reservoir .
often be the most useful way to view the
data . Some software packages can also
be used to create cross-section analysis of
DEMs, and hence, of archaeological sites
or features .
2.2 Filtering
In some areas of environmental remote
sensing, algorithms can be used to identify
the ‘signature’ of particular features, but
this has not been successfully proven
for archaeological features . Rather in
archaeology, algorithms are used to ‘clean’
the lidar data to aid visual interpretation .
In arable, pasture and moorland situations
the first or last return data on its own
is generally suitable for the recovery of
archaeological remains (indeed in open
land the first and last returns may be
identical) . Lidar comes into its own in
wooded landscapes where the use of
algorithms to filter out vegetation makes
it possible to record features beneath the
woodland canopy . It is possible in certain
circumstances to use just the last return
25
data, rather than any filtered data, but this
is very dependent on the nature of the
vegetation and on the time of flight . As
explained above, the last return data are
the result of the final return of the laser
beam from either the ground surface or
from a feature so dense that it does not
allow any of the beam to penetrate; this
may be a rock, a fallen tree trunk or in
certain cases a holly bush or other area of
dense undergrowth .
Last return data were used with great
success by English Heritage at Welshbury
hillfort in the Forest of Dean (http://www .
heritage3d .org/casestudies/2008/jan/1/
case-study-15-forest-dean/) . Here the
last pulse data revealed the bulk of the
hillfort remains, though leaving in place
off-ground ‘features’ such as tree trunks
etc . The fact that tree trunks are retained
in last return data was actually used to
assess veteran trees in Savernake Forest
where they were seen as larger ‘stumps’
than the norm . The downside to using the
last return data in wooded areas is that if
a DEM is created from it, the upstanding
tree trunks are displayed as spikes in the
model . When this is illuminated from a
low elevation to create a hill-shaded image,
the spikes show strongly, distracting the
viewer from the more subtle archaeological
features .
However, while this information can be
useful in open areas or in certain types of
woodland, it is preferable that for a fuller
and more accurate interpretation that as
many off-terrain points as possible are
removed from any dataset . The analysis
of full waveform data (see above) enables
the identification and removal of even
more off-terrain points than standard
lidar, but in practice there are always
likely to be some remaining . These are
normally readily identifiable as being of
non-archaeological origin . Because it
has always been important to be able to
create accurate DTMs for a number of
non-archaeological applications – such
as calculating topologies etc – there have
been algorithms for creating ‘bare earth’
DTMs for almost as long as there has been
access to lidar data . However, the early
filtered terrain models were not concerned
with the sort of small-scale variations that
archaeologists are usually interested in,
but were more interested in the broad lie
of the land . As a result they ran the risk
of filtering out – as noise to be removed –
those objects that the archaeologist sees as
a feature to be interpreted . Equally, and
possibly more worryingly, the resulting
surface from using these early algorithms
could also contain processing artefacts
26
that can be confused with archaeological
features; certain processes in particular
created regular gridded patterns that bear
striking similarity to ‘celtic fields’ .
These guidelines are not the
appropriate place to discuss these issues
in detail, but those wanting further details
should see Sithole and Vosselman 2004 .
Over time more sophisticated filtering/
classification methods have been devised
that create an accurate ground surface
while maintaining the subtle features in
which archaeologists are interested .
One point that should be emphasised is
that during the creation of a DTM, where
last return points are located ‘off-ground’,
options are available on how these are dealt
with . For example, they can be used in the
creation of a terrain model and the surface
forced over them (creating something
resembling a TIN) . However, this can
create false features . Alternatively, they
can be ignored and gaps left in the model
where they occur either left empty or filled
using average data from the surrounding
model . Where dense vegetation occurs,
there may be significant areas where last
returns do not reach the ground, so rather
than smoothing these areas over, there is
value in leaving them blank to emphasise
the fact that the technique was ineffective
in those areas and that further work on site
may be necessary .
2.3 Artefacts and issues
One area that needs more analysis is that
covering the various artefacts created
in lidar data . As noted above lidar data
primarily record height data, and as
such will not differentiate between
archaeological features created by human
interaction with the landscape centuries or
millennia ago and the remains of modern
agricultural or other practices . However,
as well as the features of modern origin
that need to be recognised and ignored,
there may also be some elements in the
data, and in derived images, that are not
related to any features on the ground,
but are artefacts of the original data
collection and subsequent processing .
While some of these are quite obviously
artificial, others may have an appearance
similar to archaeological features, so it is
important that these are recognised and
not misinterpreted .
One key way of recognising artefacts
is borrowed from aerial photographic
interpretation: this is to look at how a
suspect feature or pattern relates to those
features about which you have greater
confidence, for example roads and hedges .
For example, any feature that visibly
crosses a modern road or hedge, is not of
archaeological origin but is on the ‘surface’
of the image – therefore, in the case of
lidar, a data artefact . Clearly when they are
available, examination of other sources such
as aerial photographs, preferably taken at the
same time as the lidar data were captured,
will help to clarify areas of uncertainty .
There is not room in this publication to
discuss all the potential problems, but two
of the most frequently encountered and potentially misleading classes of such artefacts
can be highlighted: where the interference
patterns between overlapping swathes give
rise to wavy lines (Fig 22) and where the
interference patterns have the appearance
of ridge and furrow cultivation (Fig 33);
Fig 33 Lidar artefacts: Wave patterns (lidar © Mendip Hills AONB; source, Cambridge University ULM (April 2006)).
Fig 34 Lidar artefacts: False lynchets (lidar © Mendip Hills AONB; source, Cambridge University ULM (April 2006)).
and where rounding errors in the processing
create slight steps in the data that have the
appearance of possible lynchets (Fig 34) .
Summary
l Like an aerial photograph, a lidarderived image appears misleadingly
simple to interpret; to ensure the best
results from a survey the interpretation
should be done by someone with the
necessary skills and experience .
l There will be a significant cost benefit in
detailed evaluation of the imagery before
any field survey work .
l For predominantly non-wooded
landscapes, the possibility of
commissioning a full aerial photographic
survey using both historic and modern
photographs should be considered, as
the interpretation process is made much
easier by comparing different sources .
l Lidar has particular advantages in
wooded landscapes where the use of
algorithms to filter out vegetation makes
it possible to record features beneath the
woodland canopy .
l In certain circumstances unfiltered
last return data can reveal a significant
amount of detail beneath the canopy,
but for the best results you really need a
processed bare-earth DTM .
l As with all data sources there are
artefacts created during the original data
collection and subsequent processing;
while some of these are quite obviously
artificial, others may have an appearance
similar to archaeological features and it is
important that these are recognised and
not misinterpreted .
3 Mapping
Mapping is an essential part of
archaeological survey using lidar; in
order to adequately record the results of
interpretation of lidar data it is almost
always necessary to map the features
identified; in fact experience shows that the
actual mapping process concentrates the
mind and often clarifies the interpretation .
Depending on the level of survey and
the detail required, the same mapping
conventions as those used in aerial and
ground based archaeological surveys
can be used . Normally, mapping will be
done in a digital environment, but where
interpretation is done in the field using
paper copies of lidar imagery then the use
of manual methods on transparent overlays
may be appropriate .
Where mapping is to be carried out in
a digital environment, however, there is
a small problem . The nature of lidar and
laser-scanned data in general means that
the majority of lidar packages are designed
for viewing; they will enable data to be
seen in three dimensions, creating surface
models that can be rotated, flown through
etc . A second set of programs are available,
especially within the commercial zone, that
are designed to extract data automatically
from point clouds, such as for planning
the presence of pipes in a refinery etc .
Such programs have their uses, but the
best packages for viewing the data are
not necessarily the best for mapping and
recording purposes .
To interpret features effectively, viewing
software with full three-dimensional
functionality and controllable lighting etc
is essential, but for the mapping of the
features compromises may need to be
made . Until recently the best method was
to use the viewing software to produce a
hill-shaded raster image that could be used
as a flat base image within the mapping
software . This can be an effective method,
especially when used alongside a viewing
package that allows real-time manipulation
of the source data to aid interpretation .
New developments enable real-time
manipulation within some mapping
packages, and it is hoped that further
developments will facilitate wider use of
this technique .
The three-dimensional data can also
be used in modern photogrammetric
packages, viewed in stereo and mapped in
three dimensions . The use of such software
is still a specialised area, but may be worth
considering for particularly important sites .
It is also worth remembering that modern
digital photogrammetry can produce highresolution datasets from traditional or
modern digital photographs that are similar
to those produced by lidar; such datasets
can be used and manipulated in the same
way as lidar data and imagery .
It is important to remember that
mapping a feature or features visible on
the lidar derived imagery is only part of
the recording process; it is crucial that in
addition to the graphical depiction of any
given feature there is a database record as
well . If the mapping is carried out within
a GIS environment it is possible to attach
relevant data – such as suggested date
and interpretation – along with additional
sources, comments etc . If it is not possible
to work in a GIS package then these data
need to be recorded in a separate database
and some form of linkage made between
the two datasets .
Summary
l Mapping is an essential part of archaeological survey using lidar; in order to
adequately record the results of interpretation of lidar data it is almost always
necessary to map the features identified .
l It is important to remember that
mapping a feature or features visible on
the lidar-derived imagery is only part of
the recording process; it is crucial that in
addition to the graphical depiction of
any given feature there is a database
record as well .
4 Field use: hard copy versus digital;
raster versus vector
Lidar is still a relatively new tool in the
archaeologist’s toolkit and relatively few
projects have combined its use with field
27
survey . English Heritage survey staff have
compared the results of lidar analysis with
field survey in projects in the Mendip
Hills, in mature, deciduous woodland in
Savernake Forest, and are beginning to
do so more systematically through the
‘Miner – Farmer Landscapes of the North
Pennines Area of Outstanding Natural
Beauty’ . This early work is confirming
the accuracy and increased efficiency of
recording that the lidar data provide .
Much of the work using lidar for
archaeological investigation in the field has
so far been centred on its use in woodland,
led by the Forestry Commission . It has
been seen as a technique particularly suited
to survey in an environment in which it has
previously proved very difficult to work .
One of the key factors relating to survey
in woodland before the advent of lidar was
the issue of speed . Because of the nature
of woodland, in which features may be
obscured by the presence of trees and even
more by undergrowth, previous projects
have mostly employed ‘walkthrough’
surveys in which transects of varied
width were used (see ‘Section D: fieldbased surveys’ in Rotherham et al (eds)
2007; and concluding remarks from the
Woodland Archaeology seminar organised
in June 2003 by Gloucestershire County
Council http://www .gloucestershire .gov .
uk/index .cfm?articleid=7261) . In order to
maximise a survey, particularly given the
short timeframe during which vegetation
was at a sufficiently low level so as not
to impede study, one option was to use
large numbers of trained volunteers .
Taking this methodology forward to check
features on the ground against the lidarderived imagery, the emphasis had been
on using hard copy printouts . There are
many advantages to using such plots in
the field, especially the lack of the need
for any complex hardware or software .
A sheet of A3 paper with a hill-shaded
or PCA-composite image can provide
suitable reference material to which notes
can be added as observations are made .
Indeed this is arguably the most effective
technique even in open country .
Fig 35 Using lidar data on the GeoXt in the field.
28
2 Witham Valley
Part IV Case studies
1 Stonehenge
The Stonehenge project was the first
archaeological project in England to
make use of lidar data, albeit in a test
environment . The project was carried
out to test the usefulness of lidar and the
Stonehenge area was chosen partly because
it had been such an intensively investigated
landscape in the past . It was felt that if
lidar was able to add information in this
landscape then it was likely to be able to do
so anywhere . Unlike later projects, the lidar
data were not examined simultaneously
with the standard aerial photographs, but
rather examined separately some time
after completion of the air photo surveys .
For further details of the project and
methodology see Bewley et al 2005 .
The Witham Valley project was the first in
which English Heritage staff had access to
lidar data to manipulate . Whereas in the
Stonehenge project they could only work
with a series of flat two-dimensional images
with hill-shading, for the Witham project
they had the gridded data to manipulate .
Unfortunately the actual data available
were not as high resolution as that for the
Stonehenge survey: only data at 2m resolution (one data point for each 2m x 2m cell)
from the Environment Agency, as ASCII
gridded files . For further details of the
project and methodology see Crutchley 2006 .
Two of the key findings from the
survey were the confirmation that 2m
resolution was not necessarily high enough
for the identification of a wide range of
archaeological features, and that those
mapping the archaeological features really
required direct access to the manipulable
data, although accurate hill-shaded imagery
created specifically for archaeological
purposes would be a good second best .
As was noted in section III 1, and
Fig 36 Aerial photo of field systems south of the Winterbourne Stoke crossroads
(© English Heritage. NMR NMR21140/14 17-APR-2001).
Fig 37 Lidar imagery of field systems south of the Winterbourne Stoke crossroads
(lidar © Environment Agency 2001).
Fig 38 Unshaded lidar imagery for area around Washingborough showing a possible
barrow cemetery (centre) and the course of the Carr Dyke (left centre) (lidar courtesy of
Lincolnshire County Council; source, Environment Agency (March 2001)).
Fig 39 Colour contour and hill-shaded lidar imagery for area around Washingborough
showing the possible barrow cemetery (centre) and the course of the Carr Dyke (left centre)
(lidar courtesy of Lincolnshire County Council; source, Environment Agency (March 2001)).
especially in Figures 25 and 26, there is
a great deal of difference between the
information that can be extracted from
raw lidar imagery and from hill-shaded
imagery . Use of lidar data with greyscales
or colours depicting height, and without
hill-shading, can only reveal the most
substantial of archaeological features;
it does, however, provide a very good
topographic background to the mapping
that has its own archaeological value .
Figure 38 shows the area north-east
of Washingborough, where the National
Mapping Programme (NMP) project
recorded the remains of a possible Bronze
Age barrow cemetery . Examination of
the unshaded imagery was able to add
little in the way of confirmation of the
site but did reveal the old water channels
clearly . Analysing the data with full threedimensional capability (Fig 39) reveals the
clear outline of a number of the barrows
previously recorded only as soilmarks, as
well as delineating the course of the Carr
Dyke running from just below the middle
of the left hand edge of the image .
3 Forest of Dean
The Forest of Dean was the first time that
English Heritage staff had access to the
actual raw/ungridded lidar data and had
the necessary software to manipulate it to
extract the maximum data possible . This
was also an opportunity to look specifically
at the potential for mapping in a woodland
environment (see above) .
The project did not examine the
whole of the Forest, but rather a small
section around the Iron Age hillfort at
Welshbury . The area was surveyed at a high
resolution – up to four points per metre –
and the data provided as ungridded text
files recording x, y and z coordinates, and
intensity data for both first and last returns .
For further details of the project and
methodology see Devereux et al 2005 .
Since the survey flight, considerable
further work has been carried out in the
surrounding Forest and the benefits of
the processed data have been shown to be
enormous . Following on from both the
NMP survey and Stage 1 of the Forest
of Dean Archaeological Survey (Desk
Based Assessment), the SMR for the
Forest of Dean survey area contained
disproportionate numbers of records for
the post-medieval and modern periods
and, with the exception of the remains of
late post-medieval industrial features, the
results were heavily biased in favour of the
identification of features outside of areas of
woodland . It was as a result of this bias that
the investigation of those wooded areas was
identified as a priority for further research .
Although few of the features identified
through the lidar survey can currently
be dated with any certainty, many may
date to earlier periods previously underrepresented in the area .
Hoyle 2008 reported that preliminary
analysis identified 2,165 possible features,
of which 1,687 were thought to be archaeologically significant . The vast majority of
these had not been previously recorded, and
even where they were known sites, significant
supplementary detail was often added .
Of these 1,687 possible archaeological
sites, 46% were called significant or very
significant . They included 42 enclosures,
29
burning platforms . These were previously
known as a site type within the forest and
had been recorded in small numbers –
88 individual examples within 25 sites .
The bulk of these, however, had been
recorded in fields that were now outside
the woodland, as the recognition of such
ephemeral features within woodland had
proven difficult . The new survey increased
the number of platforms to 942 (in 111
sites) a percentage increase of 1070%
(444%), which supports the suggestion
that they may be the most common
archaeological feature within the woodland
of the Forest (Hoyle 2005, section 2 .1 .1) .
Figures 40 and 41 show the presence of a
number of previously unknown features
such as extensive field systems, together
with possible enclosures, charcoal burning
platforms and traces of more recent
mining activity .
Fig 40 DSM of the landscape west of Speech House, Forest of Dean showing little but the trees and part of the road
(lidar © Forestry Commission; source, Cambridge University Unit for Landscape Modelling (March 2004)).
Fig 41 DTM of the landscape west of Speech House, Forest of Dean revealing evidence for large scale past human activity
(lidar © Forestry Commission; source, Cambridge University Unit for Landscape Modelling (March 2004)).
some of which were interpreted as early
Roman military features or prehistoric
settlements; others are considered more
likely to be related to former field systems .
As well as these individual enclosures,
165 areas of regular linear systems, mostly
thought to be field systems, were identified;
42% of these, by area, were recorded in
woodland and covered an area of more
than 6km2 . While they are largely undated,
many are thought to be related to 13thor 14th-century assarting, while others
were almost certainly prehistoric, based
30
on their location and appearance . ‘These
features … are broadly similar in form and
give the impression of a large-scale system
of landscape organisation predating the
patterns of woodland distribution, and
[are] similar to prehistoric field systems
identified in other areas of the British Isles’
(Hoyle 2008) .’
Elsewhere, early industrial activity is
well represented by evidence of mining in
the form of small-scale surface extractive
pits and spoil tips, but the largest increase
in feature types is the recording of charcoal
4 Mendip
The Mendip Area of Outstanding Natural
Beauty AONB project was the first project
where the lidar data were used as an
integrated resource along with the standard
aerial photographs, as opposed to being used
separately after the initial interpretation had
been carried out using APs .
The landscape is mainly pasture, but
with some areas of arable cultivation and
open moorland, plus small patches of
woodland, much of it coniferous plantation .
This being the case, the lidar data provided
by the ULM were simple, unprocessed
data, as there was not felt to be the need for
the use of canopy removal algorithms .
The data were processed to provide
two outputs: one a series of georeferenced,
greyscale, hill-shaded images where height
exaggeration and controlled lighting were
used to highlight archaeological remains
to the best degree possible; and secondly
a QT file created in the Quick Terrain
Modeller format described above, which
enabled the data to be viewed interactively
in three dimensions . The greyscale images
were initially provided with a constant
illumination, but it was realised that changes
in the topography, especially the presence of
steep scarps and incised valleys, meant that
it was more useful to have individual tiles lit
in the optimum manner, whatever that might
be . The provision of the interactive QT files
meant that the features could be viewed with
illumination from all possible angles and
new hill-shaded imagery created if required .
The methodology of the project and the key
findings are recorded in Truscoe 2008 .
There were two key highlights . The first
was further data on the route of the Roman
road in the vicinity of the Roman town
Fig 42 Roman Road east of Charterhouse pre-lidar (lidar © Mendip Hills AONB; source,
Cambridge University ULM (April 2006)).
Fig 43 Roman Road east of Charterhouse post-lidar (lidar © Mendip Hills AONB; source,
Cambridge University ULM (April 2006)).
Fig 44 Ground survey of Christon medieval settlement. © English Heritage
Fig 45 Survey of Christon using lidar-derived imagery. © English Heritage.NMR.
at Charterhouse . An active local Mendip
archaeological group (Charterhouse
Environs Research Team – CHERT), had
carried out extensive area survey, including
earthwork and geophysical survey . A key
area of interest was the course of the
Roman road around Charterhouse, east
towards Old Sarum and west towards
Uphill . To the south and east, the line of
the road was recorded as an earthwork near
the Priddy Circles and again just to the
south of Ubley Warren Farm . Beyond that
point, as it headed towards Charterhouse,
the road was simply recorded as ‘course
of’, on the assumption that it was heading
towards the town by the most direct route .
CHERT members had examined the
area between Ubley Warren Farm and
the town using several survey techniques
without success . Examination of the lidar
data for the area, with a small degree
of height exaggeration and controlled
illumination, revealed the course of the
road as a low bank for over 0 .75km north
of the farm towards the Roman town (Figs
42 and 43) . Examination of the data to the
west of the town also revealed probable
traces of the road as it heads towards
Uphill near Tynings Farm .
The second highlight was working
with the EH Archaeological Survey and
Investigation team to examine the results
of lidar on the ground . The settlement and
associated field system at Christon, in the
north-west of the survey area, was chosen .
Although the area had been flown over as
part of the lidar survey and the processed
data were provided to the field team for
evaluation, it was not an area that had
been mapped as part of the NMP project
at the time of the field investigation . It
was therefore not possible to carry out a
direct comparison with the mapping and
interpretation based on the lidar data carried
out by Aerial Survey and Investigation staff .
Precisely this sort of systematic, measurable
comparison is currently being undertaken
as part of the ‘Miner–Farmer Landscapes
of the North Pennines Area of Outstanding
Natural Beauty’ project .
In the Mendips, the lidar data were
provided to the field team in the form of
hill-shaded images imported onto their
Geo-Xt hand-held GPS running FastMap
CE (see Fig 35) . The results of the
assessment suggested that while there were
some areas where the lidar data did not
31
reveal as much information as a ground
survey there were other areas where the
lidar data extended what could be seen on
the ground . In particular there were two
cases where the lidar image indicated the
continuation of several broad earthworks
beyond where they appeared to fade out
on the ground; even when the location
was recorded from the lidar data there was
nothing recognisable in the field .
After completion of the NMP mapping
for the area (Fig 44) the results from the
field survey (Fig 45) were compared with
those from the lidar interpreted imagery
and it was noted that while there was more
detail recorded in the ground survey (which
was hardly surprising given the larger
scale of the latter; see below) the bulk of the
features relating to the settlement had also
been recorded from the lidar data . That
said, some qualitative information, such
as stratigraphic relationships, can be more
easily understood from field observation
rather than from lidar data alone .
systematic examination of the archive
photography, the normal procedure for
NMP projects, also revealed a significant
number of these features . Comparing data
sources for the whole project area showed
that of 350 monuments recorded 166
(47%) were best seen on lidar, 131 (37%)
were best seen on air photos and 53 (15%)
were equally visible on both sources . For
further details of this aspect of the survey
see Crutchley 2008 .
5 Savernake
The Savernake Forest project finally
provided the opportunity to use real lidar
data in a fully interactive three-dimensional
(or rather two-and-a-half-dimensional,
as described above) environment where
mapping could take place directly onto
the data . The newly available modules
in AutoDesk Map made it possible to
insert raster surfaces into the AutoDesk
environment that enabled mapping directly
against these surfaces, and straightforward
comparison with aerial photographs .
The survey was commissioned by the
Forestry Commission so as to provide
information for their management plan
and was flown by ULM Cambridge .
The data were provided as gridded tiles
recording first return, last return and
filtered DTM showing the bare ground
surface with the maximum vegetation
removed (see above for restrictions) .
Because this was the first large-scale
project to be able to take full advantage
of the capabilities of lidar, and because
approximately half of the project area was
under woodland, one aim of the project
was to assess the relative value of using
lidar and conventional aerial photography
in a woodland environment .
When the lidar datasets were examined
a large number of features that had not
previously been recorded by either the
NMR or by the local Historic Environment
Record (HER) were noted, and it was
assumed that these were only visible
because of the capability of lidar to
penetrate the woodland canopy . In fact,
32
Fig 46 Enclosures at Church Walk as seen on lidar derived imagery (lidar © Forestry Commission; source, Cambridge
University ULM (May 2006)).
Fig 47 Enclosures at Church Walk as recorded by detailed field survey. © English Heritage.
to analyse other features that might cause
confusion during interpretation and to assess
the nature of anomalies on the ground .
The conclusion of the survey was
that comparison of the lidar image (Fig
46) and plot with the ground survey plan
(Fig 47) revealed almost exact agreement
over the location, size and shape of the
archaeological features; given the ‘soft’
nature of the earthwork detail involved,
the representations could be regarded
as ‘identical’ in terms of accuracy . The
ground survey showed a small degree of
extra detail that was not visible in the raw
lidar data, but the conclusion was that
the lidar based survey showed less detail
primarily because it was carried out at a
much smaller scale than the ground survey
(1:2500 and 1:1000, respectively) .
The quality of the lidar-derived image
was such that, even when enlarged to
1:1000, it presented a readable and useable
representation of the ground surface . It
was suggested that it was conceivable
that a suitably experienced archaeologist
could, with data of this quality, create an
accurate large-scale interpretative plan of a
site directly from the lidar-derived image,
without the need to undertake a control
survey . However, the archaeologist would
have to be aware of several limitations in
the dataset, especially: false features caused
by vegetation response and related factors;
lack of definition or absence of very slight
features (especially, on the evidence of this
study, positive ones); and discrepancies
between mapped OS detail and lidar
positioning of the same detail, exacerbated
by any enlargement of the OS base .
In this study only one relatively
minor feature was surveyed in this way .
An interesting test would be to try this
experiment by surveying a suitable site
with high quality lidar data entirely by
this method, and then check it against
electronic ground survey .
own unique set of problems for groundbased techniques, even before the advent
of lidar, and is one of the most difficult
landscapes in which to work (Bowden
1999 and Oswald et al 2008) . The arrival
of lidar with the capability to strip away
the bulk of the vegetation and reveal
the features underneath has proved to
be of great benefit (Fig 49), but it is not
without problems .
One of the key difficulties is the
fact that lidar is indiscriminate in what
it records . It has been noted that it is
important, where possible, to have an
alternative source of data to aid with
the interpretation of features . This is
especially true in woodland, particularly
in managed forestry, because there are
several practices that create features that
can easily be mistaken for archaeological
remains . Unfortunately, the nature of
woodland means that for much of it the
alternative source of aerial photography is
not available or shows nothing beyond the
top of the canopy . It is therefore doubly
important to understand the types of
features that might be seen in woodland,
and also the effects that different planting
and management regimes might have on
the results of a survey .
There are also limitations to what the
technology can show and to the types of
woodland in which it is best employed .
To gain the most from any lidar survey
commissioned for historic environment
analysis, surveys are flown at a higher
resolution than that required for open
ground, and during the winter months
when laser penetration to the forest floor
has the greatest likelihood . Many existing
lidar data may not be suitable for analysis,
as they may have been collected during
the summer and are often of a lower
resolution than the optimum required
for archaeological analysis of woodland .
Equally, if considering a new survey, it
must be emphasised that not all wooded
areas are suited to this technique .
Because the survey is dependant upon
laser penetration of the forest canopy and
understorey vegetation, significant areas
of dense, young woodland regeneration or
unthinned conifer plantation will greatly
restrict the potential of the survey and may
prevent it from being a viable option (also
see below current lidar limitations) .
The technology facilitates survey
of large areas of forest and woodland
dominated landscapes . The best results are
obtained from mature broadleaf canopy
with little understorey vegetation, for
example a beech woodland with bluebells,
where the winter survey would ensure that
the vast majority of the laser energy would
reach the forest floor uninhibited . Under
these optimum conditions, the surveys
can reveal some very subtle changes in
ground surface and reveal many subtle
archaeological features . The method is
most effective at revealing linear features
and even faint earthworks, many of which
may be difficult to see on the woodland
floor . Examples include earthworks of
enclosures (Fig 50), field systems, other
boundary banks, lynchets, route-ways
and drainage channels . When used over
Fig 48 A typical aerial photograph of a large archaeological
feature in woodland (© Peter Crow. Forest Research;
source, Cambridge University ULM (May 2006)).
Fig 49 A lidar-modelled ground surface of the same area
(© Peter Crow. Forest Research; source, Cambridge
University ULM (May 2006)).
1 Survey suitability
Part V Lidar for woodland survey
by Peter Crow, Forest Research
While aerial survey of most types of
landscape has dramatically increased our
understanding of the historic landscape,
woodland has always been a hindrance
to this process, preventing a clear view
of any archaeological evidence hidden
beneath (Fig 48) . The history of many
UK woodlands is often, therefore,
poorly understood and as such they
have been referred to as one of the UK’s
last untapped archaeological resources .
Survey in woodland also presents its
33
Fig 50 An example of a hilltop enclosure in woodland visible on lidar
(© Peter Crow. Forest Research; source, Cambridge University ULM (May 2006)).
Fig 51 Examples of circular charcoal platforms in woodland visible on lidar
(© Peter Crow. Forest Research; source, Cambridge University ULM (May 2006)).
optimum vegetation types, smaller, more
discrete features such as charcoal platforms
have been mapped (Fig 51) .
However, as noted above, lidar will not
show every historic environment feature
and will not work in all woodland types .
While the technology will work through
mature, thinned conifer – and has revealed
linear earthworks, quarries and pits under
such conditions – younger, dense conifer
plantations will greatly reduce the quantity
of energy able to penetrate to the forest
floor (Fig 52) . However, even where
canopy penetration is perceived to be good,
dense layers of understorey vegetation such
as bramble, bracken, gorse or holly can
still inhibit the laser from reaching the true
ground surface (Crow et al 2007) . Indeed
gaps in the lidar-derived DTMs caused by
understorey holly have been used to map
its distribution .
Knowledge of the vegetation types
through which the survey is expected to work
is therefore essential in considering potential
areas for lidar survey, targeting efficient use
of resources and providing confidence in the
resulting data interpretation .
While lidar has shown such discrete
features as charcoal platforms, these tend
to be several metres in diameter . However,
there is no guarantee that all platforms
of this size will be resolved, and circular
features of less than 5m diameter may
be missed . Part of the problem with the
identification of small features is that while
the lidar may have detected them, they
may only be displayed by a few pixels in
the resulting image and distinguishing
them from any noise or patches of
vegetation can be difficult .
2 Identifying features in woodland
Fig 52 Examples of the effects that different types of vegetation have on the lidar survey as shown in the
DTM (© Peter Crow. Forest Research; source, Cambridge University ULM (May 2006)).
34
As noted above, hill-shaded images will
show not only archaeological features but
will also display roads, paths, buildings
and, specifically pertinent to woodland
survey, forest residue, timber stacks and
a host of other modern objects (Fig 53) .
Additionally, changes in ground vegetation
can create patterns that look like features
of archaeological potential (Fig 54) .
Distinguishing between the genuine and
artificial historic environment is therefore
an important and necessary process,
although it is likely to be a long-term
project for survey areas of significant size .
The ability to place hill-shaded images
into GIS means that other layers can
be overlain . Placing aerial photographs,
modern and historic maps over hill-shaded
images has been discussed, but other
forest management data may also identify
many features and may provide an indirect
explanation for others . This process should
help to eliminate many objects and draw
attention to those remaining .
Where objects seen in hill-shaded
images are not identified from other
sources of information, the only reliable
method of identification is by on-site
examination . While this may not mean that
the archaeological feature or its date can
be immediately identified, it will at least
confirm that it is an earthwork or similar
structure of interest, rather than a fence or
pile of forest residue (Fig 55) .
It is likely to be impossible to ‘groundtruth’ the whole area in a short time span
and to do so would remove the value of
commissioning a lidar survey in the first
place . However, longer-term projects
may be necessary, and would require
identification of priority areas or features
for any site investigation . Professional
archaeologists may undertake this task
in conjunction with woodland managers
or it may be possible to work with local
volunteers . Indeed, there is significant
value in engaging with local groups or
communities to conduct some of this
ground-truthing . Additionally, forest staff
routinely working within the survey area
may be in the position to examine features .
Fig 53 Example of ‘false earthworks’: the bracken fallen over a wire fence can create an artificial bank
(© Peter Crow. Forest Research).
3 Lidar and managing the Historic
Environment in woodland
Important historic environment features
located within a forest need to be identified
to enable active management and to
prevent accidental damage . It is likely that
a new lidar survey will have shown a variety
of features perceived to be of historic
environment potential and interest . Unless
these features are known from other records
or site visits, it may be difficult to determine
their relative importance . Nonetheless, even
in areas where no site visit has occurred
before the commencement of operations,
the hill-shaded images can still be used
to raise awareness of potential features
and thus help forestry operations to avoid
possibly sensitive areas .
Equally, some surveys to date have
mapped landscapes of many small, but
deep pits and quarries (Fig 56) . Here the
lidar data also provide a potential map
of some on-site hazards and can be used
in conjunction with on-site assessments
to help reduce the risks of injury . This is
relevant not only to those carrying out
forest management, but also to anyone
involved in follow-up ground survey of
recorded features .
Lidar-derived data, models and indeed
any mapped features from them offer a
powerful tool for the forest design planner .
Fig 54 Example of ‘false earthworks’: this apparent mound is caused by a dense growth of rhododendron
(© Peter Crow. Forest Research).
Fig 55 Simple photographic evidence taken during routine site work can be very informative for feature identification
and management (© Peter Crow. Forest Research; lidar source, Cambridge University ULM (May 2006)).
35
Because the survey produces threedimensional surface models of a forest,
which can be manipulated within mapping
software, forest views can be examined
and planned from all angles (Fig 57) .
This has the benefit of being able to view
archaeological features as they may once
have looked in an earlier landscape, and
makes it possible for planners to consider
possible visual connections associating
historic environment features within the
landscape, or to change the setting of
individual features . Recreational access
routes to and around historic environment
features can be sensitively planned to
increase their value and profile within
the woodland, thereby enhancing its
cultural value . Indeed this is equally true
for other non-wooded environments,
such as the restoration of quarries, where
reinstatement schemes that seek to
create new wildlife and wetland habitats
incorporate new footpaths and routes for
the public . These can also take in historic
environment features that may lie just
outside (or be truncated by) a quarry . It is
then possible to have display boards that
link the newly created environment with
the original/historic environment .
Lidar data and modelled outputs
have potential uses in many areas . For
example, the differences between the
DSM and DTM can be used to produce
a map of vegetation height . The models
of the forest canopy can be used to map
individual trees, although this works best
on well-thinned or mature woodland
where there are differences in tree height
or shape due to a change in species or
establishment date . These models can
be useful in identifying and mapping
the health, structure and distributions of
ancient woodland or veteran trees within
younger plantations (Fig 58) . Hedgerows
and small areas of woodland can also be
mapped to show ecological corridors .
When a survey is carried out over mature
broadleaf woodland with little understorey,
there should be little to prevent the laser
from reaching the forest floor . Under such
conditions the large boles of any ancient
trees present (standing or fallen) can block
the laser, and thus be mapped .
With further developments in lidar
technology, it may soon be possible to map
dead wood, understorey and eventually, full
forest structure, with potential applications
in biomass calculations and carbon storage .
Lidar is a very powerful tool and
when applied to appropriate wooded
landscapes has the potential to map both
known and previously unrecorded historic
environment features . These may provide
36
Fig 56 Lidar-derived models can also be useful in mapping difficult terrain (© Peter Crow. Forest Research; lidar source,
Cambridge University ULM (May 2006)).
Fig 57 A three-dimensional model with an aerial photograph draped over it can be a very useful planner’s tool (© Peter
Crow. Forest Research; source, Cambridge University ULM (May 2006)).
information about a woodland’s history
and, in turn, guide its future management .
Nonetheless, lidar is not an instant
solution to discovering every aspect of a
woodland’s heritage and is best employed
in combination with other sources of
information . Because it is most economical
to apply the technique at the landscape
scale, costs of commissioning surveys
will inevitably be considerable . However,
such surveys should be looked upon as a
long-term investment, for the data, models
and images can be useful for planning,
management and public engagement .
Summary
l Lidar provides an unequalled means of
recording within wooded areas .
l Significant areas of dense, young
woodland regeneration or unthinned
conifer plantation will greatly restrict the
potential of the survey and may prevent
it from being a viable option .
l The best results are obtained from
mature broadleaf canopy with little
understorey vegetation .
l The method is most effective at revealing
linear features and even very subtle
earthworks can be shown, such as field
Fig 58 Shows where veteran trees have received a ‘halo-thin’ (localised management to remove competition from
surrounding trees) (© Peter Crow. Forest Research; source, Cambridge University ULM (May 2006)).
l
l
l
l
l
systems, lynchets, other boundary banks
and trackways .
Hillshaded images will not only show
archaeological features but also roads,
paths, buildings, forest residue, timber
stacks and a host of other modern
objects .
Changes in ground vegetation can
create patterns that look like features of
archaeological potential .
Hill-shaded images can be used to
raise awareness of potential features
and enable forestry operations to avoid
possibly sensitive areas .
Lidar data can also provide a potential
map of on-site hazards and can be used
in conjunction with on-site assessments
to help reduce the risks of injury .
Lidar data can be used to help planners
design recreational access routes to and
around historic environment features to
increase their value and profile within
the woodland and thereby enhance its
cultural value .
Conclusion and summary
Although lidar is a relatively wellestablished technique it has only been
used for archaeological research since
the turn of the 21st century . Because it
primarily measures three-dimensional
data it is only really effective for recording
features that exhibit some form of surface
topographic expression .
The exception to this generality is
intensity data that can be used to analyse
the reflectivity of the surface being hit by
the laser and thus in certain circumstances
may aid interpretation in a similar
way to cropmarks on traditional aerial
photographs . The accuracy and resolution
of the lidar data are heavily dependant
on the method of capture and the levels
of processing before it reaches the end
user . Standard airborne lidar is generally
said to be absolutely accurate to within
100–150mm, with a relative accuracy even
higher; it should be remembered that from
an archaeological point of view, relative
accuracy is often more important than
absolute accuracy because it is the relative
position of features that makes it possible
to record them and to understand their
relationships with other features .
When planning any sort of
archaeological survey for which it is
thought that lidar may be useful, advice
should be sought in the first instance from
the English Heritage AerSI team or from
the relevant English Heritage Regional
Science Advisor . More technical advice
may also be obtained from the Aerial
Survey and Investigation team, and the
Archaeological Survey and Investigation
team can advise on the likely cost benefits
of alternative terrestrial survey techniques .
If the survey area consists largely of
woodland, the Forest Research Team
at the Forestry Commission can provide
technical advice .
It is essential that all issues relating to
dissemination, archiving and copyright
are considered at the outset of a project
to ensure clarity regarding which data and
imagery it is possible to publish and make
available to others . Lidar data files and
generated imagery are generally quite large
and as such they are not easily supplied to
third parties .
It is important to be clear as to
whether the lidar data are required as the
primary source or whether they are seen
as a background layer for other datasets
available elsewhere . If what is required
is basic height data at scales suitable for
general topographic relief, these are also
available from alternative sources, for
example the Ordnance Survey or NASA .
If more detailed data are required it is
necessary to assess whether such data
for your area of interest already exists .
The Environment Agency have flown
large areas of the country as part of their
work monitoring flood risk, etc and a
large number of lidar surveys are carried
out by other companies each year for
non-archaeological purposes, such as
infrastructure planning etc .
One of the key factors that affect the
viability of lidar is the land use of the area to
be surveyed . Because lidar primarily records
three-dimensional data, and therefore
requires a topographic surface expression to
the features to be surveyed, the better the
earthwork survival, the better the results .
While lidar will work in most landscapes it
provides an unequalled means of recording
archaeological earthworks within wooded
areas . The best results are obtained from
mature broadleaf canopy with little
understorey vegetation, whereas significant
areas of dense, young woodland regeneration
or unthinned conifer plantation will greatly
restrict the potential of the survey and
may prevent it from being a viable option .
To achieve sufficient canopy penetration,
survey in woodland requires a higher point
density in the original data than in an open
landscape . The development of full waveform
lidar is enabling much more accurate
recoding of ground surfaces within wooded
and other heavily vegetated environments .
If new data are required then it is
advisable to look to Heritage3D, for
many of the elements with regard to
commissioning a laser scanning survey
were addressed by that project . When
considering the project area, large,
rectangular or linear survey areas are the
most cost effective, having the minimum
number of turns at the end of each
aircraft run; small or irregularly shaped
areas are the least cost effective . The
Unit for Landscape Modelling (ULM) at
Cambridge University provide a useful
calculator on their web site to assist in
planning surveys by estimating the total
flight time required .
37
38
easier by comparing different sources .
Lidar data can be used in many
formats; the standard digital product from
a lidar survey is likely to be an ASCII
grid, which can be used in standard GIS
with add-on modules or specialist threedimensional viewing programs . Ideally
the three-dimensional data should be
viewed stereoscopically, taking advantage
of the brain’s natural ability to interpret
three-dimensional objects, but if the
user does not have the facilities to view
and manipulate the original data in a
specialist package, it is still possible to use
two-dimensional snapshots of the data
as standard jpeg or Tiff files . The most
obviously user-friendly product is the
hill-shaded image that can be produced
as lit from any conceivable position .
Other visualisations, such as Principal
Component Analysis and slope models,
can also be of use .
Mapping is an essential part of
archaeological survey using lidar; in order
to adequately record the results it is almost
always necessary to map the features
identified and accompany this with a
database record . It is worthwhile noting
that even where fieldwork is intended
it can be beneficial to carry out a more
detailed desktop survey using lidar data
and other sources, such as standard aerial
photographs, and taking this information
into the field instead of, or together with,
the simple lidar derived imagery .
In summary lidar can be an extremely
useful tool when used in the appropriate
circumstances, and particularly when it
is used alongside other data sources . It is
not a magic bullet that will make possible
the recovery of all the types of features
currently recorded by other means, and
in certain cases the results will be largely
uninformative . However, when used in an
appropriate environment the results from
lidar can be spectacular .
Decision Tree
Preliminaries
2. Commissioning new data
YES
Lidar might be appropriate
YES
Lidar might be appropriate
YES
Contact Forest Research for
specialist advice
YES
Do you need data
for detailed archaeological
survey and not just for
general topography?
Is there likely to be a
surface indication of the
features you are hoping
to record?
Does your area of interest
contain a large proportion of
wooded land?
Is the woodland largely
deciduous or mixed?
You require lidar data
Is the lidar required as the
primary source, an interrogable
dataset, or is it seen as a
background layer for other
datasets available elsewhere?
YES
Consider other options as
well – OS or similar data;
US shuttle DSM; IFSAR or
LandMap for UK
Consider an alternative form
of survey, for eg. traditional air
photos, geophysics
YES
YES
Go to Section 1
YES
NO
Contact a number of suppliers
to obtain tenders etc. discuss
your requirements clearly as
they may be able to advise on
certain issues. Once you have
the data go to Section 3
YES
NO
You probably need lidar
derived imagery
Contact supplier and request
data in required format
NO
You will need to commission a
new survey – go to Section 2
1. Acquiring off the shelf data
YES
Go to Section 2
YES
Go to Section 3
Was the survey carried out to
an acceptable specification?
Eg. if for woodland, was it
carried out in leaf-off
conditions; is it at sufficiently
high resolution? (see Section 3)
Are the data available in a
suitable format; eg. is it
pre-gridded and filtered for
DTM if required; are intensity
data available?
Can the data be processed/
reverse processed to provide
it in a suitable format?
A survey may not be cost
effective; try to find additional
areas in the vicinity where
another survey can be carried
out at the same time.
If necessary, go ahead, but be
aware of high costs relative
to the area covered
NO
Look at Heritage 3D guidance
Do you know what
specification is required for
your data?
Do you have the appropriate
hardware/software to
manipulate and analyse
the data?
Do you intend to interpret
and make records of
archaeological features from
the lidar data/
lidar derived imagery?
Do you have expertise
in image interpretation?
You will need to commission a
new survey – go to Section 2
YES
Do you intend to
map archaeological features
from the lidar data/lidar
derived imagery?
NO
You will need to commission a
new survey – go to Section 2
Seek advice from someone
already using such software or
experts at English Heritage
NO
Contact supplier and request
data derived imagery
NO
You have the data, so use
them as you intended
NO
Consider contacting someone
with the appropriate skill or
approach them for advice
and/or training
NO
NO
NO
3. Using the data
YES
YES
Have you looked at the
Heritage3D guidance on threedimensional laser scanning?
NO
NO
YES
Is your area of interest covered
by a previous lidar survey?
(N.B. It’s not always easy to
find out except for
Environment Agency surveys)
Are you commissioning data
for a relatively large area
(>50km2)?
NO
Lidar penetration of coniferous
woodland, especially dense
plantations is very limited and
a survey is unlikely to produce
the results you require.
Consider an alternative form
of survey
Lidar is particularly good for
survey in deciduous woodland
YES
NO
YES
Go ahead, but remember to
utilise all other readily available
sources as well as the lidar
data to get a fully rounded
picture of the landscape you
are investigating and to minimise
the risk of misinterpreting
features seen on the lidar data
Do you have the appropriate
hardware/software to map
from the data? N.B. This is
currently quite limited
NO
Go ahead, but remember to
utilise all other readily available
sources as well as the lidar
data to get a fully rounded
picture of the landscape you
are investigating and to minimise
the risk of misinterpreting
features seen on the lidar data
NO
Seek advice from someone
already using such software or
experts at English Heritage
39
References
Barker, L, Driver, T, Murphy, K and Page,
M forthcoming 2010 ‘Monitoring coastal
erosion: remote sensing and the Iron Age
coastal promontory forts of Pembrokeshire’ .
RCAHMW Adolygiad/Review 2008/9
Bewley, R, Donoghue, D, Gaffney, V, van
Leusen, M and Wise, A 1998 Archiving
Aerial Photography and Remote Sensing
Data: A Guide to Good Practice . AHDS
Guides to Good Practice . Oxford: Oxbow;
also available online at http://ads .ahds .
ac .uk/project/goodguides/apandrs/
Bewley, R H, Crutchley, S P and Shell, C
2005 ‘New light on an ancient landscape:
lidar survey in the Stonehenge World
Heritage Site’ . Antiquity 79, 636–47
Bowden, M 1999 Unravelling the Landscape:
an Inquisitive Approach to Archaeology .
Stroud: RCHME/Tempus, 134–9
Boyd, D S and Hill, R A 2007 ‘Validation
of airborne lidar intensity values from
a forested landscape using hymap data:
preliminary analyses’ . IAPRS XXXVI, Pt 3
W52, 71–6
Challis, K 2006 ‘Airborne laser altimetry
in alleviated landscapes’ . Archaeol Prosp 13,
103–27
Challis, K, Howard, A J, Moscrop, D,
Gearey, B, Smith, D, Carey, C and
Thompson, A 2006 ‘Using airborne lidar
intensity to predict the organic preservation
of waterlogged deposits’, in S Campana
and M Forte (eds) From Space to Place: 2nd
International Conference on Remote Sensing in
Archaeology . BAR Internat Ser 1568, 93–8
Challis, K, Howard, A J, Kincey, M, and
Carey, C 2008 Analysis of the Effectiveness of
Airborne Lidar Backscattered Laser Intensity
for Predicting Organic Preservation Potential
of Waterlogged Deposits . ALSF Project 4782 .
U Birmingham
Challis, K, Kokalj, Z, Kincey, M, Moscrop,
D and Howard, A J 2008 ‘Airborne
lidar and historic environment records’ .
Antiquity 82, 1055–64
Corns A, Fenwick, J and Shaw R 2008
‘More than meets the eye’ . Archaeol Ireland
22, 3 (85) 34–8
Crow, P 2007 Historic Environment Surveys
of Woodland using lidar . Forest Research:
www .forestresearch .gov .uk/lidar
40
Crow, P, Benham, S, Devereux, B J and
Amable, G S 2007 ‘Woodland vegetation
and its implications for archaeological
survey using lidar’ . Forestry 80, 241–52
Crutchley, S P 2006 ‘Lidar in the Witham
Valley, Lincolnshire: an assessment of new
remote sensing techniques’ . Archaeol Prosp
13, 251–7
Crutchley, S 2008 ‘Ancient and modern:
combining different remote sensing
techniques to interpret historic landscapes’,
in R Lasaponara and N Masini (eds)
Remote Sensing for Archaeology and
Cultural Heritage Management; Proceedings
of the 1st International EARSeL Workshop
on ‘Remote Sensing for Archaeology and
Cultural Heritage Management’ . Rome:
ARACNE, 103–6
Höfle, B and Pfeifer, N 2007 ‘Correction
of laser scanning intensity data: data
and model-driven approaches’ . ISPRS J
Photogrammetry and Remote Sensing 62,
415–33
Holden, N, Horne, P and Bewley, R H
2002 ‘High-resolution digital airborne
mapping and archaeology, in R Bewley and
W Raczkowski (eds) Aerial Archaeology:
Developing Future Practice . 173–80
Hoyle, J P 2007 ‘The Forest of Dean,
Gloucestershire, Lidar survey of
selected areas of woodland and the
Aggregates Resource Area’ . The Forest
of Dean Archaeological Survey Stage
3A, unpublished draft report for English
Heritage
Devereux, B J, Amable, G S, Crow, P and
Cliff, A D 2005 ‘The potential of airborne
lidar for the detection of archaeological
features under woodland canopies’ .
Antiquity 79, 648–60
Jones, A F, Brewer, P A, Johnstone, E
and Macklin, M G 2007 ‘High-resolution
interpretative geomorphological mapping
of river valley environments using airborne
lidar data’ . Earth Surface Processes and
Landforms 32, 1574–92
Devereux B J, Amable, G S and Crow, P
2008 ‘Visualisation of lidar terrain models
for archaeological feature detection’ .
Antiquity 82, 470–9
Lee, E 2006 Management of Research
Projects in the Historic Environment – The
MoRPHE Project Managers Guide . Swindon:
English Heritage
Doneus, M and Briese, C 2006 ‘Fullwaveform airborne laser scanning as a
tool for archaeological reconnaissance’,
in S Campana and M Forte (eds) From
Space to Place: 2nd International Conference
on Remote Sensing in Archaeology . BAR
Internat Ser 1568, 99–105
Lennon, B and Crow, P 2009 ‘LiDAR
and its role in understanding the historic
landscape of Savernake Forest’ . Wiltshire
Archaeological and Natural History
Magazine 102, 245–61
Doneus, M, Briese, C, Fera M and Janner,
M 2008 ‘Archaeological prospection of
forested areas using full-waveform airborne
laser scanning’ . J Archaeol Sci 35, 882–93
English Heritage 2003 ‘Where on Earth are
We? The Global Positioning System (GPS)
in archaeological field survey’ . Swindon:
English Heritage
English Heritage 2007 ‘Understanding
the Archaeology of Landscapes: a guide to
good recording practice’ . Swindon: English
Heritage
Graham, L, 2007 ‘LAS specification
version 2 .0 (final draft)’ . March 7, 2007,
ASPRS Lidar Committee
Hampton, J N 1974 ‘An experiment
in multispectral air photography
for archaeological research’ . The
Photogrammetric Record 8, 37–64
Loisios, D, Tzelepis, N and Nakos, B 2007
‘A methodology for creating analytical hillshading by combining different lighting
directions’ . Proceedings of 23rd International
Cartographic Conference, Moscow, August
2007 or http://cartography .tuwien .ac .at/ica/
documents/ICC_proceedings/ICC2007/
documents/doc/THEME%2020/Oral%20
2/A%20METHOLOGY%20FOR%20
CREATING%20ANALYTICAL%20
HILL-SHADING%20BY%20
COMBININ .doc
McCue, J, Millard, K, von Lany, P and
Clark, M 2004 ‘Position Review of Data
and Information Issues within Flood and
Coastal Defence’ . R&D Technical Report
FD2314/PR . Bristol: Environment Agency
Oswald, A, Smith, L and Handley, C
2008 ‘Introduction to surveys and survey
techniques’, in I D Rotheram, M Jones, L
Smith, and C Handley Woodland Heritage
Manual – A Guide to Investigating Woodland
Landscapes . Sheffield: Wildtrack, 86–95
Page, M, Barker, L, Driver, T and Murphy,
K forthcoming 2010 ‘Remote sensing and
the Iron Age coastal promontory forts of
Pembrokeshire’ . Archaeol Wales 47
Pfeifer, N and Briese, C 2007 ‘Geometrical
aspects of airborne laser scanning and
terrestrial laser scanning’ . ISPRS Workshop
Laser Scanning 2007, Espoo, Finland, 0912-2007–09-14-2007 . IAPRS XXXVI, Pt
3 W52, 311–19
Sithole, G and Vosselman, G 2004
‘Experimental comparison of filtering
algorithms for bare-earth extraction from
air-borne laser scanning point clouds’ .
ISPRS J Photogrammetry and Remote
Sensing 59, 85–101
Sorenson, G P, Honey, R C, and Payne,
J R 1966 ‘Analysis of the use of airborne
laser radar for submarine detection and
ranging’ . SRI Report 5583, Stanford
Research Institute (CONFIDENTIAL)
Stone, J L and Clowes, M 2004
‘Photogrammetric recording of the Roman
earthworks ‘‘Cawthorn camps’’, North
Yorkshire’ The Photogrammetric Record 19,
94–110
Truscoe, K 2008 ‘Archaeological aerial
survey in the Central Mendip Hills .
The Aggregate landscape of Somerset:
predicting the archaeological resource’ .
Aggregates Levy Sustainability Fund
project 3994, unpublished report:
Somerset County Council/English
Heritage
Verhoeven, G and Loenders, J 2006
‘Looking through black-tinted glasses – a
remotely controlled infrared eye in the sky,
in S Campana and M Forte (eds) From
Space to Place: 2nd International Conference
on Remote Sensing in Archaeology . BAR
Internat Ser 1568, 73–9
Wehr, A and Lohr, U 1999 ‘Airborne laser
scanning an introduction and overview’ .
ISPRS J Photogrammetry and Remote
Sensing 54, 68–82
Glossary
algorithm A step-by-step problem-solving
procedure, especially an established,
recursive computational procedure for
solving a problem in a finite number of steps .
ALS (Airborne Laser Scanning) Lidar is
actually a generic term for all forms of laser
measuring, whether ground based or aerial,
and so ALS can sometimes be a better
acronym .
cell One value in a raster that corresponds
to a specific point or area, often referred
to as a pixel . A raster cell value may be the
elevation above sea level at one position
in a survey site or the intensity of red
radiation for a pixel in a video image .
DEM (Digital Elevation Model) A grid
of cells or of pixels with a height value
assigned to each square . This type of grid
is often called a raster . It is differentiated
from a standard raster image in that the
value assigned to each cell is a height value
rather than a tonal one . This is the broad
term that encompasses both the DSM and
DTM .
DSM (Digital Surface Model) This is a
digital elevation model of the land surface .
It records the highest points including
buildings and the woodland canopy . In
lidar terms this is generated by the first
return of the laser pulse .
DTM (Digital Terrain Model) This is a
digital elevation model of the bare earth –
ie the ground beneath any vegetation with
other structures such as buildings removed .
In lidar terms this is generated by filtering
the last return of the laser pulse using an
algorithm to calculate where features
exist above the natural ground surface and
removing them .
equalisation An image processing
technique that redistributes the brightness
values of the pixels in an image, so that
they more evenly represent the entire range
of brightness levels . Equalisation remaps
pixel values in the composite image so
that the brightest value represents white,
the darkest value represents black, and
intermediate values are evenly distributed
throughout the greyscale .
first pulse/return The first echo of the
laser pulse . The laser pulse is sent out
from the sensor towards the ground .
When any part of the footprint hits a
reflective object part of it is returned to
the sensor . The first object struck provides
the first pulse or first return . In open
ground there is often only a single return,
but any form of vegetation will produce
multiple returns .
footprint The footprint of the laser beam
is the area covered by the diverging beam
when it strikes a surface .
full waveform The more recent form
of laser recording that instead of just
recording between two and four returns
digitises the entire analogue echo waveform
for each emitted laser beam . During
post-processing the full waveform can
be modelled, for example as a series of
Gaussian distribution functions, each
representing an interaction between
individual objects and the laser .
geoid A mathematical model of the level
surface closest to the mean sea level over
the oceans . The surface is continued
under the land and acts as a fundamental
reference surface for height measurement .
GIS (Geographical Information System)
A Geographical Information System is an
information system for capturing, storing,
analyzing, managing and presenting data
that are spatially referenced .
GPS (Global Positioning System) A
satellite-based positioning receiver and
navigation system that pinpoints the
geographic location of the user to differing
degrees of accuracy, depending on the
equipment .
grid A geographic representation of the
world as an array of equally sized square
cells arranged in rows and columns;
each grid cell is referenced by its x and y
locations .
HER (Historic Environment Record)
Historic Environment Records are the
mainly local-authority-based services used
for planning, but they also operate a public
service and play a role in education . These
records were previously known as Sites and
Monuments Records or SMRs: the name
has changed to reflect the wider scope of
the data they now contain .
hill-shade The hypothetical illumination
of a surface . A hill-shade raster can be
calculated for a given surface or hillshading can be applied on the fly . A hillshaded image is a computer-generated
image used to show subtle changes in the
topography of DEMs by the use of shadow
in the same way that subtle earthworks
can be highlighted by low-angled winter
sunlight . An artificial sun position is
defined and used to illuminate the DEM .
IFSAR (Interferometric Synthetic
Aperture Radar) By combining the
principals of Synthetic Aperture Radar
(SAR) with Interferometry, Interferometric
Synthetic Aperture Radar (IFSAR) is
41
capable of producing both a radar image
of the ground surface and calculating
elevation changes to enable production of a
digital surface model (DSM) . IFSAR data
are available in the form of a 5m spatial
resolution DSM with a vertical accuracy
of between 0 .5m and 1 .0m, and a 1 .25m
spatial resolution radar image . The low
spatial resolution of the IFSAR data means
that although it is able to distinguish broad
geomorphological zones, such as the river
terraces and floodplains, it is of limited
value for archaeological purposes .
IMU (Inertial Measurement Unit) An IMU
works by sensing its own rate and direction
of motion using a combination of accelerometers and gyroscopes, which then enable
a guidance computer to track its position
using a process known as dead reckoning .
interferometry The use of interference
phenomena between a reference wave
and an experimental wave or between two
waves to determine wavelengths and wave
velocities, measure very small distances
and thicknesses, and calculate indices of
refraction . Radar interferometry relies
on picking up the returned radar signal
using antennas at two different locations .
Each antenna collects data independently,
although the information they receive
is almost identical, with little separation
(parallax) between the two radar images . The
phase difference between the signals received
by each of the two antennas is used as a basis
for the calculation of changes in elevation .
intensity The strength of the signal
returned to the sensor . As well as the time
taken to return to the sensor that helps
calculate the physical location of the point
on the ground, the sensor also records the
strength of the returning signal . This gives
some indication of the reflectance of the
surface struck by the beam; rough surfaces
generally return weaker signals as part of
the beam is dispersed and reflected away
from the sensor .
interpolation The process of inserting,
estimating or finding a value intermediate
to the values of two or more known points
in space . In the case of lidar data this
generally relates to the estimation of an
elevation value at an unsampled point
based on the known elevation values of
surrounding points .
42
last pulse or last return This is the last
echo of the laser pulse . The laser pulse
is sent out from the sensor towards the
ground; the last return is the final echo
returned to the sensor . In the majority of
cases in open land this will represent the
ground surface, but it can also represent
extremely dense vegetation that no part
of the beam can penetrate . It can also
represent any solid surface above ground
level, such as a building .
lidar (Light Detection and Ranging)
Variously written as lidar, Lidar and
LIDAR . Also known as ALS .
nodes The point at which areas (lines,
chains, strings) in a polygon network are
joined . Nodes carry information about the
topology of the polygons .
orthophoto An orthophoto or
orthophotograph is an aerial photograph
that has been geometrically corrected
(‘orthorectified’) such that the scale is
uniform: the photo has the same lack of
distortion as a map .
panchromatic a greyscale representation
of all the visible wavelengths .
PCA (Principal Component Analysis)
A multivariate statistical technique to
structure complex datasets . In the course
of investigating lidar data, it can be used
to examine many hill-shaded images
and compile a composite image showing
more than 95% of the variations seen
within them all .
photogrammetry Photogrammetry is
the process of obtaining reliable
information about physical objects and
the environment through processes of
recording, measuring and interpreting
photographic images . Specifically,
photogrammetric packages make it
possible to map and interpret visible data
in three dimensions .
pixel see cell
point cloud The raw data format from
the lidar survey, comprising millions of
x, y and z coordinates in the form of
text . Some software packages make
it possible to view this data as threedimensional points .
raster A grid of data used within GIS
software . For elevation models, the cells
hold height data; for hill-shaded images,
the cells hold tonal values .
SAR (Synthetic Aperture Radar) A radar
system in which a series of microwave
pulses are emitted continuously at a
frequency constant enough to be coherent
for a fixed period; all echoes returned
during this period can then be processed as
if a single antenna as long as the flight path
had been used .
SMR (Sites and Monuments Record) The
former name for HERs; still in use in some
parts of the country .
TIN (Triangular Irregular Network) A
data structure that represents a continuous
surface through a series of irregularly
spaced points with values that describe the
surface at that point (eg their elevation) .
From these points a network of linked
triangles of varying sizes forms the surface .
This is a key difference to a raster surface,
for which the grid is regular .
three-dimensional visualisation In
the context of these guidelines threedimensional refers to the ability to
visualise the height element of the data
in a meaningful way . This means that it
is possible to create images where the
elevation data can be coded and even
exaggerated so as to produce imagery with
shadows and highlights, in either oblique
or plan view . This is sometimes referred
to as 2½-dimensional to differentiate
it from true three-dimensional viewing
(stereoscopic viewing), which requires
even more specialised equipment, such as
monitors or polarising glasses .
vegetation removal A computer-based
process to filter out data from the point
cloud derived from vegetation, making it
possible to create a DTM
WGS84 ‘WGS’ stands for the World
Geodetic System and the ‘84’ comes
from the fact that the latest revision dates
from 1984 . This is a standard for use in
cartography, geodesy and navigation, and
comprises a standard coordinate frame for
the earth, a standard spheroidal reference
surface for raw altitude data and a geoid
that defines the nominal sea level .
Appendix
Sources of advice on using lidar
English Heritage
Within English Heritage the first point
of contact for general archaeological
science enquiries, including those relating
to using lidar data, should be the English
Heritage regional science advisors, who
can provide independent, non-commercial
advice. Such advisors are based either in
universities or in the English Heritage
regional offices. Please contact regional
advisors currently based in universities at
their university address.
North West
Sue Stallibrass
University of Liverpool
Department of Archaeology, Classics and
Egyptology (SACE)
Hartley Building, Brownlow Street
Liverpool L69 3GS
telephone: 0151 794 5046
fax: 0151 794 5057
[email protected]
North East
Jacqui Huntley
Department of Archaeology
University of Durham
Science Laboratories
Durham DH1 3LE
telephone and fax: 0191 33 41137
[email protected]
Yorkshire
Andy Hammon
English Heritage regional office
37 Tanner Row
York YO1 6WP
telephone:01904 601 983
fax: 01904 601 999
mobile: 07747 486255
[email protected]
West Midlands
Lisa Moffett
English Heritage regional office
8th Floor, The Axis
10 Holiday Street
Birmingham B1 1TG
telephone: 0121 626 6875
mobile: 07769 960022
[email protected]
South East
Dominique de Moulins
Institute of Archaeology
UCL Room 204
31–34 Gordon Square
London WC1H 0PY
telephone: 020 7679 1539
fax: 020 7383 2572
[email protected]
East Midlands
Jim Williams
English Heritage regional office
44 Derngate
Northampton NN1 1UH
telephone: 01604 735451
mobile: 07801 213300
[email protected]
Specific advice on the use of lidar data for
archaeological research can be sought from
Simon Crutchley in the English Heritage
Aerial Survey and Investigation team.
East of England
Jen Heathcote
English Heritage regional office
24 Brooklands Avenue
Cambridge CB2 8BU
telephone: 01223 582759
mobile: 07979 206699
[email protected]
South West
Vanessa Straker
English Heritage regional office
29 Queen Square
Bristol BS1 4ND
telephone: 0117 975 0689
[email protected]
Simon Crutchley
English Heritage
Kemble Drive
Swindon SN2 2GZ
telephone: 01973 414704
[email protected]
Forestry Commission
For advice on the use of lidar data for
archaeological research in woodland
contact:
Peter Crow
Forest Research
Alice Holt Lodge
Wrecclesham
Farnham
Surrey
GU10 4LH
London
Dr Rachel Ballantyne
c/o Greater London Archaeology
Advisory Service
1 Waterhouse Square
138-142 Holborn
London EC1N 2NH
43
Written by Simon Crutchley with a special contribution on using
lidar in woodland by Peter Crow, Forest Research, Forestry
Commission . These guidelines should be cited in bibliographies
and references as follows: Crutchley, S and Crow P 2009 The
Light Fantastic: Using airborne laser scanning in archaeological
survey . Swindon: English Heritage .
These guidelines are available via the English Heritage Free
Publications list at www .english-heritage .org .uk/publications
Acknowledgments
Pete Horne, Mark Bowden
and Trevor Pearson
www .forestresearch .gov .uk/lidar
Thanks also to Michael Doneus, David Barber, Mark Kincey and
Sarah Jackson who also read and commented on earlier drafts;
and finally, to John Vallender, who prepared the illustrations .
The Aerial Survey and Investigation team of English
Heritage provides a greater understanding of England’s historic
environment through aerial reconnaissance, combined with
interpretation, mapping and analysis of information contained
on aerial photographs and related sources . We work closely with
many of English Heritage’s specialist teams, particularly with the
Archaeological Survey and Investigation teams .
We also work with many partners outside English Heritage,
from national organisations to local amateur groups, and provide
regular work placements for university students and professionals
wishing to learn more about the work of the team . All our
work helps to expand and update the database of information
available to everyone through English Heritage’s public archive,
the National Monuments Record and its web enabled access
Pastscape, and you can obtain copies of any of our photographs,
reports and publications from there .
Published April 2010
Edited and brought to press by David M Jones,
English Heritage Publishing
Designed by www .mark-simmons .com
Printed by [to be filled in by printer]
If you would like this document in a different format,
please contact our Customer Services department:
Telephone: 0870 333 1181
Fax: 01793 414926
Textphone: 01793 414878
customers@english-heritage .org .uk
Minimum of 75% recovered fibre, the remainder
being from sustainable sources .
Product Code 51577
44
Front cover: Savernake Forest near Marlborough,Wiltshire. Three
views of the forest with a traditional aerial photograph, first return lidar
data DSM showing the tree canopy and processed DTM revealing the
archaeology hidden beneath the trees (NMR 21339/19 (10-AUG2001) © English Heritage; lidar image © Forestry Commission, from
Cambridge University ULM (May 2006)).
Back cover: Previously unrecorded settlement of probable Late Iron
Age/Romano-British date with associated boundaries, hollow ways and
lynchets near Alston, Cumbria revealed by lidar (© English Heritage,
from Infoterra (May 2009)).
English Heritage is the Government’s statutory advisor on
the historic environment . English Heritage provides expert
advice to the Government about matters relating to the historic
environment and its conservation .
For further information and copies of this publication, quoting
the Product Code, please contact:
English Heritage, Customer Services Department
PO Box 569, Swindon SN2 2YP
telephone: 0970 333 1181
customer@english-heritage .org .uk
If you would like this document in a different format, please contact
our Customer Services department:
Telephone: 0870 333 1181
Fax: 01793 414926
Textphone: 01793 414878
E-mail: [email protected]